JPH0658725A - Detecting method of attribute of high-luminance area - Google Patents

Abstract

PURPOSE:To enable detection of the position of a slit light image on an image by a method wherein the width of a peak area being near to a base level of a luminance peak and higher than a threshold is converted into a value of an index being small, medium or large, a slit-shaped reflection corresponding to the obtained value of the index is calculated and the attitude of the peak area is detected. CONSTITUTION:An image signal obtained by picking up the image of an object 3 by a camera 2 is converted into digital data. An image memory 4 stores processed data, while a display unit 5 and a printer 6 output the result of processing of a computer 1 and it is registered in a floppy disk 7. The computer 1 calculates the position of a slit light on the object 3, develops an image of the central position in the image memory 4 with a higher resolution than the resolution of the camera 2 and makes the display unit 5 display it, while transferring this image of the central position to a host computer. Based on this image of the central position, the host computer calculates the shape and dimensions of the object 3.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the detection of the properties of a slit light reflection image on an image obtained by shooting a slit light projection object with a television camera. In particular, although not intended to be limited to this, slit position detection is possible. In the measurement of the object shape, the brightness peak area on the image is "slit shape" that accurately represents the slit light reflection image, or "blooming" or smear that causes an error in the slit position detection. The present invention relates to a method for detecting and accurately measuring an object shape.

[0002]

2. Description of the Related Art For example, laser light is projected in a slit shape on an object, the object is photographed by a television camera, and the slit light reflection position on the object on the photographed screen is calculated by image processing and calculation by a computer. However, the measuring method for calculating the shape, size, etc. of the object is convenient because the object can be measured online and in real time without contact. In this measurement, it is necessary to accurately measure the high-intensity line by the slit light and the line width center position of the line on the object. Japanese Unexamined Patent Publication No. 2-95091 discloses S of image data.
N-ratio, amount of break of high-intensity line due to slit light, and
Roughly, there is a technique for determining the appropriateness of the intensity of slit light and the binarization threshold value of the image signal, that is, the appropriateness of the measurement environment or the like, by fuzzy reasoning from the line width of the high-intensity line. Proposed.

[0003]

However, the above-mentioned S
Accurate values cannot be obtained for any of the N ratio, the amount of break, and the line width. For example, due to the unevenness of the object itself, the break of the high-intensity line due to the slit light appears on the image.
This break is one parameter for calculating the shape and size of the object. Therefore, it is necessary to identify whether the break of the high-intensity line is due to the shape of the object or the break due to an inappropriate measurement environment. The line width is also the same. In the calculation of the SN ratio, which of the image data is the noise N causes a problem to be passed on to the reliability of noise detection. As described above, the detections are required one after another, but each detection has a problem of reliability. Therefore, there is no evidence that a highly reliable result can be obtained.

In addition, blooming or smear may appear on the image. This is shown in FIG. Bull-
The ming BL is a phenomenon in which the image signal of the camera is saturated to the maximum level due to high-brightness reflection due to a peculiar high-reflection spot on the object.
Due to the characteristics, a high level signal appears at the corresponding position of another scanning line, and this is the smear SM. That is, blooming B
L causes smear SM. FIG. 14 shows the luminance signal level in one scanning line of the (normal) region NL that accurately represents the slit light image, FIG. 15 shows the luminance signal level in one scanning line of the blooming BL, and one scanning line of the smear SM. FIG. 16 shows the luminance signal level at. by the way,
It is important to detect the edge of the slit light image on the object (the boundary between the low-luminance area and the high-luminance area: each of the width edges of the slit light image), and the edge is especially important when the shape and dimensions are required accurately. Accurate determination of the center position of (line) is required. By the way, for example, a two-dimensional camera in which 512 photoelectric conversion elements (pixels) are arrayed in both the x and y directions (for example, C
In photographing using a CD camera), when the pixel array pitch is Kp and the camera magnification is Kc, the pixel array pitch Kp on the pixel array plane represents the length of Kp · Kc of the actual object. Since it is not possible to obtain a resolution less than the pixel array pitch Kp on the pixel array surface, the resolution (minimum unit) in measurement is Kp · Kc. Therefore, when the magnification of the camera is high or the pixel array pitch Kp is large, the measurement accuracy is low. In addition to this, there is a problem in that the edge cutout (edge extraction) accuracy is reduced due to the edge line becoming blurred or thick depending on the brightness of the illumination, the light reflectance of the object surface, and the like. Explaining this in a little more detail, as shown in FIG. 2B, the object 3 is photographed by the CCD camera 2 as shown in FIG. 1 to obtain an image frame as shown in FIG. When the region to be processed is extracted, the luminance signal (digital) of one scanning line y = j becomes as shown in FIG. 3A, and the luminance signal sharply changes at the edge portion of the slit light image. The edge detection position changes depending on the level of the threshold value for cutting out (extracting) the peak area (slit light image area). In addition, the edge detection position changes due to a slight change in the brightness of the illumination light or the reflected light.

The first object of the invention of the present application is to detect the attribute of the peak area such as whether the peak area on the image is a normal slit light image, blooming, smear, or the like. The second purpose is to accurately extract the slit light image on the image, and the third purpose is to accurately detect the image edge of the slit light image on the image. The fourth purpose is to detect accurately.

[0006]

In a first embodiment (claim 1) of the present invention, an image data representing the brightness of each pixel distributed in two-dimensional x, y is obtained by photographing an object illuminated with slit light. The attribute of the high-luminance area on the image represented by the data is detected as follows.

A. Luminance peak (Lm on scan line (y = j)
x) is detected, B ₁ . Detecting a first peak region width (LSW) on the scanning line that is equal to or higher than a first threshold (0.2Lmx) close to the base level of the luminance peak, which is determined by the detected luminance peak (Lmx), D ₁ . The first peak area width (LSW) is small, medium, large, etc. [fL
z (LSW) to fLb (LSW)] values [fLz (d), fLs (d)], and E ₁ . The obtained index values [fLz (d), fLs (d)] are associated with each index.
Converted to the value of each higher index (fNz (fLz) ~ fNb (fLb)), the center value (NLv) of the set of these values, that is, the "slit shape"
The certainty factor (NLv) that is a reflection is calculated.

In the second embodiment (claim 2), A. Detecting a luminance peak (Lmx) on the scan line (y = j), C.I. Brightness peak - click (Lmx) detects the saturation width (MSW) when a luminance saturation value (255), D _2. Saturation width (MSW) is small, medium,
Values of each index (fMz (MSW) ~ fMm (MSW)) (fMz (c), fMs
It converted to (c)], and, E _2. Each obtained index value [fMz (c), fMs (c)] is associated with each index.
Converted to the value of each higher index (fBz (fMz) ~ fBm (fMm)), the central value (BLv) of the set of these values or `` blooming ''
The certainty factor (BLv) that is a reflection is calculated.

In the third embodiment (claim 3), A. Detect the luminance peak (Lmx) on the scanning line (y = j), B ₁ . Detecting a first peak region width (LSW) on the scan line that is equal to or higher than a first threshold (0.2Lmx) close to the base level of the luminance peak, which is determined by the detected luminance peak (Lmx), B ₂ ． Determined by the click (Lmx), the luminance peak - - detected luminance peak second threshold close to click (0.8Lmx) above, the second peak of the scanning line - detecting the click region width (USW), D _3. 1st peak area width (LSW) and 2nd peak area width (USW)
Relative value (e = USW / LSW) of small, medium, large index [fEz (E) ~ f
Converting the value of em (E)] [fEs (e) ~fEm (e)], and, E _3. Each index value obtained [fEs (e) ~ fEm (e)] is associated with each index, and the value of each index [fSz (fEz) ~ fSm (fEm)] is low to some extent in "smear". And the confidence value (SMv) which is the central value (SMv) or “smear” of the set of these values.
Calculate v).

In the fourth embodiment (claim 4), as in the first embodiment, the certainty factor (NL) which is a "slit shape" reflection.
v), the confidence (BLv) that is a “blooming” reflection is calculated in the same manner as in the second embodiment, and F ₁ ..
By comparing the certainty factor (NLv) that is a "slit shape" reflection and the certainty factor (BLv) that is a "blooming" reflection, the peak area is defined as the "slit shape" reflection when the former (NLv) is large. , The latter (BLv) is large, it is determined to be a "blooming" reflection.

In the fifth embodiment (claim 5), as in the first embodiment, the certainty factor (NL
v) is calculated, and the certainty factor (BLv) that is the “blooming” reflection is calculated in the same manner as in the second embodiment, and the certainty factor (SMv) that is the “smear” is calculated in the same manner as in the third embodiment. Calculate and F ₂ . Confidence (NLv) that is a "slit-shaped" reflection,
Among the certainty (BLv) that is the “blooming” reflection and the certainty (SMv) that is the “smear”, the attribute of the maximum value, that is, the “slit shape” reflection, the “blooming” reflection, or the “smear” Is determined as the attribute of the peak area.

In a preferred embodiment of the present invention, the center of each of the edge range where the brightness rises and the edge range where the brightness falls falls in the peak area determined as the "slit shape", and the center between the calculated centers is calculated. Is set as the center of the peak area, and when there is a scanning line determined as "blooming", the intensity of the slit light applied to the object is reduced and the image is taken again. For the scanning line determined to be "," the attribute of the scanning line is detected again, and the center of each of the edge range in which the brightness rises and the edge range in which the brightness falls in the peak area determined to be the "slit shape" is determined. The center between the calculated centers is determined as the center of the peak area, and the peak area determined as "smear" is excluded from the target for determining the center of the peak area.
(Claims 6 to 8).

In a preferred embodiment of the present invention, the center of the edge range is calculated as follows (claims 9 to 9).
11).

A. Calculate a difference in luminance between adjacent pixels, that is, a differential value (Di) in the direction (x direction) in which the scanning line (y = j) extends, b. The differential value (Di) changes from 0 to another value and returns to 0. The pixel of the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2) within the edge range (xeb1 ~ xee1 / xeb2 ~ xee2) in the direction in which the scanning line extends Convert each differential value (Di) to the value of each index such as small, medium, large (Fig. 6
A), c ₁ . Pixels (xi) of the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2)
For each of the above, each index value is converted into each edge existing width according to the edge existing width conversion function associated with each index, and the first combined central value which is the combined central value of each edge existing width is calculated. (A in FIG. 7), c ₂ . Pixels (xi) of the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2)
For each of the above, each index value is converted into each edge probability according to the edge probability conversion function associated with each index, and the second combined central value that is the combined central value of each edge probability is calculated (FIG. 7). of b), and, d _3. Pixels (Di) in the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2)
Position (xi) in the direction in which each of the scanning lines (y = j) extends is weighted by the first and second combined center values, and the scanning line in the direction in which the scanning line in the predetermined region formed by a group of these pixels extends. Calculate the center position (Ejcp1 / Ejcp2).

Symbols in parentheses represent drawings showing corresponding items or processing modes shown in the drawings.

[0016]

When the slit light image normally appears on the image, that is, when it has a "slit shape" on the image,
As shown in FIG. 14, the luminance distribution on one scanning line shows a shape close to an upwardly convex parabola, but in the case of blooming, as shown in FIG. 15, there is a region where the luminance is saturated, The peak area width becomes wider. In the case of smear,
As shown in FIG. 16, the luminance distribution is close to a triangular shape, and the peak area width is narrow. That is, the luminance distribution during blooming has a wide saturation region width (MSW) and therefore a wide peak region width, and is close to the base level of the luminance peak, which is determined by the luminance peak (Lmx). First peak area width (LSW) above the threshold (0.2Lmx) and luminance peak (Lmx)
Since the second peak region width (USW) equal to or larger than the second threshold (0.8Lmx) close to the luminance peak (Lmx) of the luminance peak, which is determined by, is wide, the relative value E = USW / LSW is large. On the contrary, smear has substantially no saturation region (saturation region width = 0), and
= USW / LSW is extremely small. In the case of the normal "slit shape", the luminance distribution shows a property almost intermediate between those of blooming and smear. Therefore, using at least one of the first peak region width (LSW), the saturation region width (MSW) and the relative value E = USW / LSW of the first and second peak region widths, the luminance distribution is calculated. It is possible to discriminate whether the peak area is a "slit shape" characteristic (normal), a "blooming" characteristic, or a "smear" characteristic.

In the first embodiment (claim 1), the first peak region width (LSW) is set to each of small, medium and large indexes [fLz (LSW) to fLb (LSW).
W)) value [fLz (d), fLs (d)], and obtained each index value [f
Lz (d), fLs (d)] is associated with each index, and each is a "slit-shaped" reflection with low and high indices [fNz (fLz)
~ FNb (fLb)], and the median of the set of these values
(NLv), that is, the certainty factor (NLv) that is a "slit-shaped" reflection
Therefore, the larger the certainty factor (NLv), the higher the probability that the peak area on the luminance distribution has the “slit shape” (normal). In other words, the lower the certainty factor (NLv), the higher the probability that the peak area is "blooming" or "smear". In this way, the evaluation value (NLv) indicating the degree to which the peak area on the luminance distribution has the "slit shape" (normal) is obtained.

Similarly, according to the second embodiment (claim 2), an evaluation value (BLv) indicating the degree to which the peak area on the luminance distribution is "blooming" (abnormal) is obtained, Third aspect
According to (Claim 3), the evaluation value (SMv) indicating the degree of "smear" (abnormal) can be obtained.

In the fourth embodiment (claim 4), the certainty factor (NLv) which is the "slit-shaped" reflection and the certainty factor (BLv) which is the "blooming" reflection are compared to determine the peak area. , The former (NL
If v) is large, it is determined as "slit-shaped" reflection, and if the latter (BLv) is large, it is determined as "blooming" reflection.
The probability of detecting "Ming" as "slit shape" (normal) is lower.

In the fifth embodiment (claim 5), the certainty factor (NLv) which is a "slit shape" reflection, the certainty factor (BLv) which is a "blooming" reflection and the certainty factor (SMv) which is a "smear". )
Among these, the attribute of the maximum value, that is, "slit shape" reflection, "blooming" reflection or "smear" is determined as the attribute of the peak area, so "blooming" or "smear" is determined. The probability of detecting the "slit shape" (normal) is the lowest, and the information indicating whether or not the peak area on the luminance distribution has the "slit shape" (normal) with the highest reliability.

In a preferred embodiment of the present invention, the center of each of the edge range where the brightness rises and the edge range where the brightness falls falls in the peak area determined as the "slit shape", and the center between the calculated centers is calculated. Is set as the center of the peak area, and when there is a scanning line determined as "blooming", the intensity of the slit light applied to the object is reduced and the image is taken again. For the scanning line determined to be "," the attribute of the scanning line is detected again, and the center of each of the edge range in which the brightness rises and the edge range in which the brightness falls in the peak area determined to be the "slit shape" is determined. The center between the calculated centers is determined as the center of the peak area, and the peak area determined as "smear" is excluded from the target for determining the center of the peak area.
(Claims 6 to 8). According to this, the position of each slit light image on each scanning line is accurately detected without wasting each captured image, and substantially without leakage and without picking up noise (smear). Shape measurement is possible.

In the preferred embodiment of the present invention, the calculation of the center of the edge range is performed by a. Calculating a difference in luminance between adjacent pixels, that is, a differential value (Di) in a direction (x direction) in which the scanning line (y = j) extends, b. The differential value (Di) changes from 0 to another value and returns to 0. The pixel of the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2) within the edge range (xeb1 ~ xee1 / xeb2 ~ xee2) in the direction in which the scanning line extends Convert each differential value (Di) to the value of each index such as small, medium, large (see Fig. 6
a), c ₁ . Pixels (xi) of the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2)
For each of the above, each index value is converted into each edge existing width according to the edge existing width conversion function associated with each index, and a first combined central value that is a combined central value of each edge existing width is calculated. (A in FIG. 7), c ₂ . The predetermined area (xeb1 to xee1 / x
For each pixel (xi) of eb2 to xee2), each index value is converted into each edge probability according to the edge probability conversion function associated with each index, and the second is a composite central value of each edge probability. Calculate the composite center value (b in FIG. 7), and d ₃ . Pixels (Di) in the predetermined area (xeb1 ~ xee1 / xeb2 ~ xee2)
Position (xi) in the direction in which each of the scanning lines (y = j) extends is weighted by the first and second combined center values, and the scanning line in the direction in which the scanning line in the predetermined region formed by a group of these pixels extends. The central position (Ejcp1 / Ejcp2) is calculated (claims 9 to 11).

This a. Differentiating process of, each pixel in a predetermined area within the range from the differential value changing from 0 to another value and returning to 0
The differential value of (xi) is converted into a value representing the degree of belonging to each index such as small, medium, large, etc. that vaguely represents the brightness.
Processing, the converted value is converted into an edge existence width and an edge probability according to the edge existence width conversion function and the edge probability conversion function associated with each index, and the composite center value and edge probability of the edge existence width are converted. Of the above-mentioned c ₁ . , C ₂ . Processing, and the position of each pixel (xi) of the predetermined region, are weighted with these synthetic center value, x-direction center position or edge center of the area - position (Ejcp1, Ejcp2) calculates a d _3. By the processing of step 1, the numerical value representing the edge center position (Ejcp1, Ejcp2) includes a fractional value expressed in the pixel unit (pixel pitch), and the edge center information (Ejcp1 , Ejcp2) is obtained.

In addition, the differential value Di has little influence of the brightness of the illumination, and based on this, the above-mentioned c ₁ . , C ₂ . And d ₂ . Processing, especially d ₂ . Since the center value of the set does not substantially move due to fluctuations in the brightness of the illumination, the obtained edge center information (Ejcp1, Ejcp2) has no detection error due to the illumination.

Other objects and features of the present invention will become apparent from the following description of embodiments with reference to the drawings.

[0026]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the configuration of an image processing apparatus for carrying out the present invention in one mode. This device comprises a computer 1, an image pickup camera 2, an image memory 4, a display 5, a printer 6, a floppy disk 7, a keyboard terminal 8,
A laser light source 14 for projecting a slit light 14se onto the object 3,
A laser driver 1 for energizing this light source to adjust the emission intensity
It will be 5 mag. An x-axis traveling carriage 9 is mounted on the y-axis traveling carriage 11, and the y-axis traveling carriage 11 is driven in the y direction by an electric motor 12 via a power transmission mechanism (not shown). The electric motor 12 is a motor driver 13
Positive and reverse rotation is activated by y. The x-axis traveling carriage 9 is driven in the y direction by an electric motor 10 via a power transmission mechanism (not shown). The electric motor 10 is a motor driver 13x
It is urged forward and backward. The object 3 to be measured is fixed on the x-axis traveling carriage 9, and the object 3 is slit-shaped.
The light (hereinafter, slit light) 14se is projected. Camera 2
Takes an image of the upper surface of the object 3.

The camera 2 shoots the object 3 to obtain 512 × 5.
This is an ITV camera (CCD camera) that outputs an analog image signal in 12 pixel sections, and the A / D converter 22 converts this image signal into digital data (image data) of 256 gradations per pixel. . The image memory 4 shown in FIG. 1 is readable and writable and stores various processing data including the original image data of the captured image. The display unit 5 and the printer 6 output the processing results of the computer 1 and the like, and the floppy disk 7 registers the processing results. Further, the keyboard terminal 8 is operated by an operator to input various instructions or data. A host computer is further connected to the computer 1, and in accordance with an instruction or data given from the host computer or an instruction or data given from the keyboard terminal 8, the computer is connected. -Ta 1 photographs the object 3, and the slit light SLi (FIG. 2) position (S) on the object 3 is photographed on the screen.
The center position image (thin line slit image) is developed in the image memory 4 at a resolution higher than that of the camera 2 and displayed on the display 5, and the center position is calculated. Transfer the image to the host computer. The host computer calculates the shape and size of the object 3 based on the center position image. In addition,
1 and 2 show a mode in which the hole shape and the size of the upper end surface of the object 3 are measured.

FIG. 9 shows the outline of the slit light reflection position measuring process of the computer 1. Light reflection position measurement L
In PD, the computer 1 first sets the light emission intensity of the light source 14 (subroutine IDP. In the following, in parentheses, the word subroutine or step is omitted and only the symbol or number is described.). The light emission tonality is adjusted by a subroutine 15 (FIG. 10) which will be described later. In this laser power setting IDP, the past laser power adjustment 15 (is set to the reference value set by the program or by external input. The light emission intensity of the light source 14 is set to the value (adjustment allowance) adjusted in FIG. 10) plus a correction value with a higher weight from the latest one in terms of time.

Next, the computer 1 uses the A / D converter 22 to output a video signal for one screen of the image picked up by the camera 2 from 0 to 0.
It is converted into 255-gradation image data and written in the original image area (X, Y coordinates) of the memory 4 (1). An image of the written original image area is shown in FIG. The computer 1 then determines a predetermined processing area of the original image area (four corners indicated by a chain double-dashed line inside the outer four corners of FIG. 2A: X = x ₀ , Y = y ₀ and X = x).
Area where ₀ + a, Y = y ₀ + b is a diagonal point = the image data of (b) in FIG. 2 is extracted and the processed image data memory area (hereinafter, extracted data area) of the memory 4 is extracted. X,
This is the y coordinate, and the origin 0, 0 is X = of the original image area.
Write to x ₀ , Y = y ₀ ) (2).

Next, the j-th image data of the extraction data area
Differentiate the image data of the line (line of y = j) (4). j = 1 is the line to be processed first. The differentiation process is performed in the x direction (from left to right in FIG.
Is subtracted from the image data I _i-1 of the pixel x _{i-1 one} pixel before (on the left side) of the image data I _i (luminance digital value). That is, the differential value Di = Ii−I of the target pixel xi
_i-1 . As shown in (b) of FIG. 2, since the background has low luminance and the slit light image SLi on the object 3 has high luminance, the image luminance (digital) of one line is as shown in (a) of FIG. In the edge region that is distributed and changes from the background (the surface of the object 3) to the slit light image SLi and vice versa, the image data greatly changes. The horizontal axis of (a) of FIG. 3 represents the arrangement of pixels (one pixel on one scale), and the vertical axis represents the luminance of the pixel (one unit of 0 to 255 gradations on one scale).

In contrast to such image data distribution, the differential value Di becomes the distribution shown in FIG. 3B, and the differential value Di changes from 0 to positive at the place where it enters the first edge regions xeb1 to xee1. Then, when entering the second edge area xeb2 to xee2, the differential value Di
Changes to a negative value, and the differential value Di returns to 0 at the place where the second edge region exits. The first edge areas xeb1 to xee1 + the second edge areas xeb2 to xee2 correspond to the slit light image SLi.
Area (Fig. 3). Focusing on such a change, the computer 1 checks the distribution of differential values for one line continuous in the x direction, and determines the peak area (first edge area xeb).
1 to xee1 + second edge area xeb2 to xee2) is detected (5). When the peak area is detected, the computer 1 determines whether the peak area is a "slit-shaped" reflection (normal) in which the slit light image accurately appears, or a "blooming" reflection. Whether or not it is "smear" (6), if it is judged as "slit shape" reflection (normal), "NL" is set in the register Pch, and if it is judged as "blooming" reflection, it is set in the register Pch. BL ”for“ smear ”
If it is determined that "SM" is written in the register Pch (characteristic determination 6 of the peak area). Details of this content will be described later with reference to FIG. 11.

The computer 1 then determines (P) that it is a "slit-shaped" reflection (normal) according to the above determination result.
ch = “NL”), the center positions (Ejcp1, Ejcp2) of the first and second edge regions (xeb1 to xee1 and xeb2 to xee2 in FIG. 3) are calculated, and the line No. The center position calculated by the correspondence is written in the thinned data memory area of the memory 4 (7, 8). Details of this content will be described later with reference to FIG.

Judgment as "blooming" reflection (Pch =
If it is "BL"), "1" is written in the register BFR to indicate that there is blooming, and the line No. is written in the table BFR (allocated to the memory 4) for storing blooming information. .j and the brightness saturation width MSW calculated in the property determination 6 of the peak area are written (7, 9). If the determination is “smear” (Pch = “SM”), the peak
Area is ignored. That is, no processing is executed on the peak area (7 → 10). Every time one peak area is detected, the characteristic judgment (6) of the peak area and the processing (7 to 8) corresponding to the judgment are executed, and when the processing of the j-th line is completed, the target line Is updated to the next line (12), and the differentiation processing (4) of that line is performed, and similarly, the peak area detection (5) and the following are performed (5-1).
0).

Referring to FIG. Final line (y = j =
When the process of b) is completed, it is checked whether the content of the register BFR is "1" (there is blooming) (13), and the content is "0" (there is no blooming). ) And the center position data of each line of the thinned data memory area (the respective center positions E of the first edge areas xeb1 to xee1 and the second edge areas xeb2 to xee2 in FIG. 3).
The center value [(Ejcp1 + Ejcp2) / 2] of jcp1, Ejcp2) is calculated, and the image (center line image of the slit light image) showing this position is developed in the high-resolution image memory of the display 5 and displayed on the display 5. And the central value [(Ejcp1 + E
jcp2) / 2] data is transferred to the host computer, and the image data in the high-resolution image memory is output to the display 5, printer 6, floppy disk 7 and / or host computer (19). ). The host computer sequentially accumulates the transferred data, draws the three-dimensional shape of the hole on the upper end surface of the object 3, and calculates the shape and size of the hole. In order to obtain stereoscopic information, the host computer gives the computer 1 information for instructing various positions of the carriage 9, and in response thereto, the computer 1 moves the carriage 9 to the instructed position. Establish, FIG. 9 and FIG.
The "light reflection position measurement" LPD shown in 0 is executed. Now,
If the content of the register BFR is "1" (there was blooming) when the processing of the final line (y = j = b) is completed, the table BFR data for storing the blooming information will be deleted.
Search the highest saturation width MSW in the data (14). Then, the light emission intensity of the light source 14 is reduced corresponding to the saturation width MSW (15). Then, the above "image input"
Similarly to (1) and "extraction of image area" (2), the image data of one screen of the camera 2 is read and a predetermined area (b in FIG. 2) in the image data is extracted (16). Then, the data of the table BFR for storing the blooming information is saved as the evasion table.
Move to the table (memory) PDT, and clear the register BFR and the table BFR (18). Then, for each of the lines written in the table PDT (the line determined to be blooming in the previous measurement on the shooting screen),
The processes (3 to 12) below the line differentiation process described above are similarly executed (3a to 12a in FIG. 10). Since the intensity of the slit light is reduced, the number of lines determined to be blooming is small or does not exist in the captured shadow image this time. Therefore, this processing (3a to 1) is performed.
2a), the center position of the peak area (slit light image) of the line, which could not be measured due to blooming in the previously captured image, is measured, and this measurement data is obtained in the previously captured image. It will be added to the measurement data. When blooming is detected in the line which is the second measurement with the slit light intensity lowered, step 11 in FIG.
From step a to step 13, the second "laser power adjustment" (15) is executed, and the second measurement process (3a to 1).
2a) is executed. In this way, the adjustment of the slit light intensity and the re-measurement (3a to 12a) of the line in which the blooming is detected are executed until the blooming is no longer detected, and the re-measurement is performed without detecting the blooming. When the measurement is completed, the measurement of all the lines in which there is blooming in the first photographing is completed without blooming, and the measurement data obtained for all times (substantially shown in b of FIG. 2). (Data of all lines in the area), that is, the center position data of the slit light image
(Center positions Ejcp1 and Ej of the first edge regions xeb1 to xee1 and the second edge regions xeb2 to xee2 of FIG. 3 respectively).
jcp2) exists in the thinned data memory area. The computer 1 calculates the center value [(Ejcp1 + Ejcp2) / 2] of the center positions Ejcp1 and Ejcp2, and displays the image (center line image of the slit light image) indicating this position on the display 5
The image data of the high-resolution image memory is transferred to the host computer while the central value [(Ejcp1 + Ejcp2) / 2] data is transferred to the host computer. Output to the display 5, printer 6, floppy disk 7 and / or host computer (1
9). Next, referring to FIG. 11, the contents of the “characteristic determination of the peak area” (6) shown in FIG. 9 (the contents of 6a in FIG. 10 are the same) will be described. First, the luminance peak value Lmx of the peak area (first edge area xeb1 to xee1 + second edge area xeb1 to xee1 in FIG. 3) is detected (6
1). From this value Lmx, the first threshold value 0.2Lmx is calculated, and the number of pixels in the peak area whose brightness level is equal to or higher than the first threshold value d =
The LSW (peak base width) is calculated (62). Next, the second threshold value 0.8Lmx is calculated from the luminance peak value Lmx and the peak value is calculated.
Number of pixels in the black area with a brightness level above the second threshold Usw
(Peak width) is calculated (63), and the relative value e = USW / LSW of these calculated values is calculated (64). Further, it is checked whether or not the luminance peak value Lmx is a saturation value of 255 or more, and if it is, the number of pixels c = MSW of the saturation value of 255 or more in the peak area is counted. When the luminance peak value Lmx is less than the saturation value 255, the numerical value 0 is assigned to c = MSW (65).

The computer 1 then determines the peak base width d =
The degree value fLz (fLz (LSW), fLz (LSW), fLm (LSW) and fLb (LSW) belonging to the respective indices fLz (LSW), fLz (LSW), fLz (LSW) of zero (z), small (s), medium (m) and large (b) d), fLs (d), fLm
(d) and fLb (d) (66). The conversion function is shown in FIG. In this example, since d = LSW is small, both fLm (d) and fLb (d) are 0 (zero).
Is.

Then, the "slit shape" is associated with the index (z, s in the illustrated example) that has obtained a value exceeding 0.
The index values fLz (d) and fLs (d) are converted into contribution values to the certainty factor according to a weighting function of each index with respect to the certainty factor considered as reflection (normal), and a combined value NLv of these contribution values is calculated. (67). FIG. 5A shows the functions fNz (fLz), fNs (fLs), fNm (fLm), and fNb (fLb) used for this calculation.
These are functions that define the area. For example, the degree value fLz (d) belonging to the index (z) is the function fNz (fLz)
It is converted into the horizontal axis position of the center of gravity of the area of the area defined by the horizontal axis and having the vertical axis value fLz (d) or less. The degree value fLs (d) belonging to the index (s) is the horizontal axis position of the center of gravity of the area of the figure defined by the function fNs (fLs) and the vertical and horizontal axes below the vertical axis value fLs (d). To be converted. The combined center of gravity of these positions of the horizontal axis is NLv. The horizontal axis represents the extent to which the peak area accurately represents the slit light image. That is, it represents the certainty factor that is "slit-shaped" reflection (normal). In NLv, the peak area to be processed is "slit-shaped" reflection (normal)
Is the certainty factor.

The computer 1 then outputs the saturation width c = MSW.
Are zero (z), small (s) and medium (m) indices
Degree value fM belonging to fMz (MSW), fMs (MSW) and fMm (MSW)
Convert to z (c), fMs (c) and fMm (c) (68).
The conversion function is shown in FIG. In this illustrated example, c =
Since MSW is slightly large, fMz (c) is 0 (zero).

Then, "blooming" is associated with the index (s, m in the illustrated example) that has obtained a value exceeding 0.
The index values fMs (c) and fMm (c) are converted into contribution values to the certainty factor according to the weighting function of each index with respect to the certainty factor regarded as (abnormal), and the combined value BLv of these contribution values is calculated ( 69). FIG. 5B shows the functions fBz (fMz), fBs (fMs) and fBm (fMm) used for this calculation. These are functions that define the area. For example, the degree value fMs (c) belonging to the index (s) is the horizontal axis of the center of gravity of the area of the figure defined by the function fBs (fMs) and the vertical and horizontal axes, and below the vertical axis value fMs (c). Converted to position. Degree value fMm (c) belonging to index (m)
Is converted into the horizontal axis position of the center of gravity of the area of the figure defined by the function fBm (fMm) and the vertical and horizontal axes and having the vertical axis value fMm (c) or less. BLv is the composite center of gravity of these center of gravity horizontal axis positions. The horizontal axis represents the probability that the peak area is blooming. That is, it represents the certainty factor that is "slit-shaped" reflection (normal). In BLv, the peak area to be processed is
The certainty factor is "blooming" (abnormal).

The computer 1 then selects the peak of the peak area.
Ratio of peak width USM to peak bottom width LSW e = E = U
SW / LSW is zero (z), small (s) and medium (m)
Are converted into degree values fEz (e), fEs (e) and fEm (e) belonging to the respective indexes fEz (E), fEs (E) and fEm (E) (70). The conversion function is shown in FIG. In this example, since e = USW / LSW is slightly large, fMz (c)
Is 0 (zero).

Subsequently, the index value fEs is calculated according to the weighting function of each index with respect to the certainty factor that is regarded as "smear" (noise), which is associated with the index (s, m in the illustrated example) that has obtained a value exceeding 0. (e) and fEm (e) are converted into contribution values to the certainty factor, and a combined value SMv of these contribution values is calculated (71). The function fS used for this calculation is shown in FIG.
z (fEz), fSs (fEs) and fSm (fEm) are shown. These are functions that define the area. For example, the degree value fSs (e) belonging to the index (s) is the horizontal axis of the center of gravity of the area of the region defined by the function fSs (fEs) and the vertical and horizontal axes and having the vertical axis value fSs (e) or less. Converted to position. The degree value fSm (e) belonging to the index (m) is
The vertical axis value fS of the figure defined by the function fSm (fEm) and the vertical and horizontal axes
It is converted into the horizontal axis position of the center of gravity of the area of the area of m (e) or less. The combined center of gravity of these center of gravity positions on the horizontal axis is SMv.
The horizontal axis represents the probability that the peak area is smear. That is, it represents the certainty factor that is "smear" (noise). SM
v is the peak area to be processed is "smear" (noise)
Is the certainty factor.

As described above, the peak region is the certainty factor NLv that is the "slit-shaped" reflection that accurately represents the slit-shaped image, and the certainty factor BLv and "smear" that are the "blooming". When the certainty factor SMv is calculated, the computer 1 compares NLv, BLv and SMv, and determines the highest value among them (72 to 74). When NLv is the largest, NL is set in the register Pch (the peak area is "slit shape"), and when BLv is the largest, BL is set in the register Pch (the peak area is "blooming").
If SMv is the largest, SM is added to the register Pch.
(The peak area is "smear"), that is, the information indicating the determination result is written (75 to 77).

Next, referring to FIG. 12, "edge center E
Calculation of jcp ”(8) will be described. Roughly speaking, this calculation (8) is performed by the center position Ejcp1 (b in FIG. 8) of the first edge regions xeb1 to xee1 of the peak region shown in FIG. 3 and the x of the second edge regions xeb2 to xee2. Direction center position Ejc
It is for calculating p2. (Ejcp1 + Ejcp2) / 2 is
It is the center position of the peak region in the x direction, that is, the center position of the slit light on the object 3 in the x direction.

In the calculation (8) of the center position Ejcp, the computer 1 first determines the pixel of interest as the pixel xeb1 at the beginning (left end) of the first edge region (81a), and the pixel xi =
The degree value fd (z = Di), in which the differential value Di (absolute value) of xeb1 belongs to each index of zero (z), small (s), medium (m), large (b) and extra large (u), fd (s = Di), fd (m = D
i), fd (b = Di) and fd (u = Di) (82a).
The conversion function is shown in FIG. In this illustrated example, Di
= 3, fd (m = Di), fd (b = Di) and fd (u = D
Both i) are 0 (zero). Next, the computer 1 follows the index value [fd (z = Di, according to the edge existence width (wz, ws) estimation function associated with the index (z, s in the illustrated example) that has obtained a value exceeding 0. ), Fd (s = Di)] is converted into an edge existence width (estimated value: wz, ws in a of FIG. 7), and a composite edge existence width (Wwi) of the obtained edge existence width is calculated (83a). . The function f used for this calculation is shown in FIG.
w (z), fw (s), fw (m), and fw (b) are shown. These are functions that define the area. For example, the degree value fd (z = Di) belonging to the index (z) is the area of the region of the figure defined by the function fw (z) and the vertical and horizontal axes and having the vertical axis value fd (z = Di) or less. It is converted to the horizontal axis position wz of the center of gravity. Degree value fd (s = D belonging to index (s)
i) is converted to the horizontal axis position ws of the center of gravity of the area of the area defined by the function fw (s) and the vertical and horizontal axes and having the vertical axis value fd (s = Di) or less. The composite center of gravity of these center of gravity positions wz and ws is Wwi. The horizontal axis is the estimated width of edge existence,
Indicates how much the edge area extends from the pixel of interest xi.

Next, the computer 1 follows the index value [fd (z in accordance with the edge likelihood estimation function, that is, the edge probability estimation function associated with the index (z, s in the illustrated example) that has obtained a value exceeding 0. = Di), fd (s = Di)] into edge probabilities (wz, ws in FIG. 7B), and a composite edge probability (Wai) of the obtained edge probabilities is calculated (84a). FIG. 7B shows functions fa (z), fa (s), fa used for this calculation.
(m) and fa (b) are shown. These are functions that define the area. For example, the degree value fd (z = Di) belonging to the index (z)
Is converted into the horizontal axis position wz of the center of gravity of the area of the area defined by the function fa (z) and the vertical and horizontal axes and having the vertical axis value fd (z = Di) or less. The degree value fd (s = Di) belonging to the index (s) is the vertical axis value fd (s = D of the figure defined by the function fa (s) and the vertical and horizontal axes.
i) Converted to the horizontal axis position ws of the center of gravity of the area of the following area.
The composite center of gravity of these center of gravity positions wz and ws is Wai. The horizontal axis represents the edge probability estimated value, and Wai represents the probability that the target pixel xi is the edge center.

In this way, the computer 1 can detect the area x
The edge existence width estimated value Wwi and the edge probability Wai are calculated for all pixels xi of eb1 to xee1 (82a to 86).
a). Next, the computer 1 uses the first edge area xeb1 ...
For each pixel xi of xee1, a triangle in which the spread Wwi of the edge region is taken in the x direction (horizontal axis) around the x position of the pixel xi and the edge center-probability Wai is taken as the vertical axis (one per pixel) On the same x position (horizontal axis) and edge probability (vertical axis; edge likelihood) plane (FIG. 8).
The solid line triangle (a) and the two-dot chain line triangle), the outer contour line of all the developed triangles, and the x position Ejcp1 of the center of gravity of the figure surrounded by the horizontal axis (b in FIG. 8) are calculated. This position Ejcp1
Is the center (x direction) of the first edge regions xeb1 to xee1, that is, the center position in the x direction.

Next, the computer 1 performs the same processing as the calculation of the center position Ejcp1 of the first edge area (81a to 87a) described above, and the center position in the x direction of the second edge area xeb2 to xee2 (FIG. 3). Ejcp2 is calculated (81b to 87b).
When this x-position data Ejcp2 is calculated, computer 1
Is a line N in the thinned data memory area of the memory 4.
o. Write the x position data Ejcp1 and Ejcp2 to the register assigned to j (88). With the above, the line No.
The center positions (x direction) of the first edge regions xeb1 to xee1 and the second edge regions xeb2 to xee2 of j are obtained.

By the way, in the edge region, the change of the image data (luminance data) is large, and the absolute value of the differential value Di (luminance difference between adjacent pixels in the x and y directions) is large, and the differential value D
The position where i (which should be interpreted as an absolute value below) is the highest (hereinafter x
However, there is a high probability that it will be the edge center. From this viewpoint, the degree value fd (z = Di) belonging to each index of zero (z), small (s), medium (m), large (b) and extra large (u) shown in (a) of FIG. fd (s = D
i), fd (m = Di), fd (b = Di) and fd (u = Di) are index values fd (u = Di) from the index (u) indicating that the larger the differential value Di is, The index functions, that is, the membership functions fd (z), fd (s), fd (m), fd (b) and fd (u) are determined so as to be larger, and the edge existence width in FIG. In the estimation functions fw (z), fw (s), fw (m) and fw (b), the edge position is considered to be nearer to the pixel as the differential value Di is larger, and thus the small edge existence width estimated value (Center of gravity w
z, ws, etc.) should be given. On the other hand, the edge probability (edgelikeness) estimation functions fw (z), fa (z), in FIG.
Fa (s), fa (m) and fa (b) are given larger edge probability estimation values (center of gravity wz, ws, etc.) in order to give greater weight to the position of the pixel as the differential value Di is larger. Stipulated in. The centroid Wwi of the set of edge existence width estimation values (centroids wz, ws, etc.) shown in (a) of FIG. 7 is a first weighting value assuming that the position xi of the pixel xi is the edge center. As shown in (b) of FIG. 7, the center of gravity Wai of the set of edge probability estimated values (center of gravity wz, ws, etc.) has a position xi of the pixel xi at the edge center.
Is a second weighting value. These values Ww
One pixel xi of FIG. 8A defined by i and Wai
The center of gravity of the addressed triangle (solid line) is the combined value of the first and second weighting values Wwi and Wai, that is, the edge area xeb1.
It becomes the weighting value of one pixel xi among all the pixels of xee1. Since there is one edge center, the barycentric position of such a set of weighting values for all the pixels can be determined as the edge center. In the above embodiment, the horizontal axis position of the center of gravity of the figure surrounded by the outer contour line (b in FIG. 8) and the x-axis of the set of triangles (solid line & two-dot chain line) addressed to each pixel in FIG. The x position Ejcp1 is determined to be the edge center.

[0048]

As described above, according to the present invention, the high-brightness region on the screen where the object illuminated by the slit light is photographed is
Is it a "slit shape" that accurately shows the slit light image,
Attributes such as "blooming" with saturated brightness due to high spots, or "smear" due to afterimage of the camera are detected. The reliability of object shape measurement by light projection can be increased. The attribute detection is highly reliable because it is based on the shape characteristics of the high-brightness peak area appearing on the screen by the slit light irradiation.

Therefore, the center position detection of the slit light image based on this attribute detection is highly reliable and accurate, and the center position data of a resolution higher than that of the camera is detected.
You can get the data.

[Brief description of drawings]

FIG. 1 is a block diagram showing a device configuration for implementing the present invention in one aspect.

2A is a plan view showing a screen imaged by the camera 2 shown in FIG. 1, and FIG. 2B is a graph showing a luminance distribution of a scanning line having a processing region of the screen.

3A is a graph showing a luminance distribution of a scanning line j shown in FIG. 2B, and FIG. 3B is a graph showing a luminance difference between adjacent pixels, that is, a luminance differential value.

4A is a graph showing a luminance distribution of the scanning line j shown in FIG. 2B, and FIG. 4B is a peak lower width L shown in FIG.
A graph showing a function for converting SW into an index value showing the degree of the size, (c) a graph showing a function for converting the brightness saturation width MSW shown in (a) into an index value showing the degree of the size, And, (d) is a graph showing a function for converting the ratio E of the peak upper width USW to the peak lower width shown in (a) into an index value indicating the degree of the size.

5A is a graph showing a weighting function for converting an index value obtained by the function shown in FIG. 4B into a certainty factor of “slit shape”, and FIG. 5B is shown in FIG. The graph which shows the weighting function which converts the index value obtained by the function shown to the certainty factor of "Blooming", and (c) shows the index value obtained by the function shown in (d) of FIG. It is a graph which shows the weighting function which converts into a certainty factor.

FIG. 6A is a graph showing a function for converting a luminance differential value into an index value indicating a degree of magnitude, and the luminance of the first edge region and the first edge region of the luminance distribution shown in FIG. Shown together with the differential value.

7A is a graph showing an edge existence width estimation function for converting a specified value obtained by the function shown in FIG. 6A into a certain weight value, and FIG. 7B is a graph showing another index value. It is a graph which shows the edge probability estimation function converted into a weight value.

8A is an edge existence width estimated value calculated using the estimation function shown in FIG. 7A and an edge probability estimated value calculated using the estimation function shown in FIG. 7B. 6 is a graph showing the weight of the certainty of estimation that the defined pixel of interest xi is the edge center, which is indicated by a solid line triangle. (B)
[Fig. 6] is a graph showing an outer contour line of a set of weights of estimation certainty that each pixel in the edge existence region is an edge center.

FIG. 9 is a flowchart showing a part of the processing contents of the computer 1 shown in FIG.

FIG. 10 is a flowchart showing a part of the processing contents of the computer 1 shown in FIG.

11 is a "characteristic determination of the peak area" shown in FIG.
It is a flowchart showing the contents of (6).

FIG. 12 “Calculation of edge center Ejcp” shown in FIG.
It is a flowchart showing the contents of (8).

FIG. 13 is a plan view showing an outline of an image with the camera 2 shown in FIG. 1, showing blooming and smear.

FIG. 14 is a graph showing a luminance distribution of an image area shown as “normal” (NL) in FIG.

FIG. 15 is a graph showing a luminance distribution of an image area shown as blooming (BL) in FIG.

16 is a graph showing the luminance distribution of the image area shown as smear (SM) in FIG.

[Explanation of symbols]

1: Computer 2: Television camera 2f: Photographing screen 3: Object 3i: Object image SLi: Slit light image 4: Image memory 5: CRT display 6: Printer 7: Floppy disk 8: Keyboard 9: x moving carriage 10: motor 11: y moving carriage 12: motor 13x: motor driver 13y: motor driver 14: laser light source 14se: slit light 15: laser driver 16: slit exposure / imaging Device 22: A / D converter

Claims (11)

[Claims] 1. A high brightness area of an image represented by image data representing the brightness of each pixel distributed in a two-dimensional x, y image of an object irradiated with slit light, which comprises the following steps: Attribute detection method: A. Brightness peak on the scanning line - detecting the click, B _1. Detected luminance peak - determined by click, the luminance peak - first threshold value or more first peak close to the basal level of click - detecting the click region width, D _1. The first peak - the click region width small, medium, and converts a large etc. of the value of each index, and, E _1. Each obtained index value is associated with each index,
The "slit-shaped" reflection is converted to values of low and high indices to some extent, and the central value of the set of these values, that is, the certainty factor of the "slit-shaped" reflection is calculated. 2. A high-luminance area of an image represented by image data representing the luminance of each pixel distributed in two-dimensional x, y obtained by photographing an object illuminated by slit light, which comprises the following steps. Attribute detection method: A. Detecting a luminance peak on a scan line, C.I. Brightness peak - click detects the saturation width when a luminance saturation value, D _2. Convert the saturation width into values for small, medium, large, etc. indices, and E ₂ . Each obtained index value is associated with each index,
"Blooming" reflection is converted into values of low and high indices to some extent, and the central value of the set of these values, that is, "blue-
Compute the confidence that is the "Ming" reflection. 3. A high-luminance region of an image represented by image data representing the luminance of each pixel distributed in two-dimensional x, y obtained by photographing an object illuminated by slit light, which comprises the following steps. Attribute detection method: A. Brightness peak on the scanning line - detecting the click, B _1. Determined by the click, the luminance peak - - detected luminance peak above the first threshold value close to the basal level of click, the first peak of the scanning line - detecting the click region width, B _2. Detected luminance peak - determined by click, the luminance peak - at least a second threshold value close to the click, the second peak of the scanning line - detecting the click region width, D _3. The first peak - click region width and a second peak - the relative value of the click region width small, medium, and converts a large etc. of the value of each index, and, E _3. Each obtained index value is associated with each index,
"Smear" is converted to low and high index values to some extent,
The certainty factor that is the central value of the set of these values, that is, "smear" is calculated. 4. A high-luminance area of an image represented by image data representing the luminance of each pixel distributed in two-dimensional x, y obtained by photographing an object illuminated by slit light, which comprises the following steps. Attribute detection method: A. Brightness peak on the scanning line - detecting the click, B _1. Detecting a first peak region width equal to or higher than a first threshold, which is determined by the detected luminance peak and is close to the base level of the luminance peak, C. When the luminance peak is the luminance saturation value, the saturation width, which is the width thereof, is detected, D ₁ . The first peak - the click region width small, medium, and converts a large etc. of the value of each index, D _2. Convert the saturation width to the value of each index such as small, medium, large, etc. E ₁ . D ₁ . Each index value obtained in step 1 is converted into the value of each of the low, high, etc. indexes with the "slit shape" reflection, which is associated with each index, and the central value of the set of these values, that is, the "slit shape" reflection Calculate certain confidence, E ₂ . Above D ₂ . Each index value obtained in step 1 is converted into a value of each index, which is associated with each index and has a certain degree of low or high reflection, and the central value of the set of these values, that is, "the blooming reflection". Is calculated, and F ₁ .. By comparing the certainty factor of "slit-shaped" reflection and the certainty factor of "blooming" reflection, the peak area is larger when the former is "slit-shaped" reflection, and when the latter is larger, "blooming". It is decided to be a reflection. 5. A high-luminance area of an image represented by image data representing the luminance of each pixel distributed in two-dimensional x, y obtained by photographing an object illuminated with slit light, which comprises the following steps. Attribute detection method: A. Brightness peak on the scanning line - detecting the click, B _1. Detected luminance peak - determined by click, the luminance peak - first threshold value or more first peak close to the basal level of click - detecting the click region width, B _2. Detecting a second peak area width on the scanning line that is equal to or larger than a second threshold value close to the brightness peak, which is determined by the detected brightness peak, C. When the luminance peak is the luminance saturation value, the saturation width, which is the width thereof, is detected, D ₁ . The first peak - the click region width small, medium, and converts a large etc. of the value of each index, D _2. Converting the saturation width small, medium, to a value greater such each index, D _3. Converting the relative value of the first peak area width and the second peak area width to the value of each index such as small, medium, and large, E ₁ . D ₁ . Each index value obtained in step 1 is converted into the value of each of the low, high, etc. indexes with the "slit shape" reflection, which is associated with each index, and the central value of the set of these values, that is, the "slit shape" reflection Calculate certain confidence, E ₂ . Above D ₂ . Each index value obtained in 1. is converted into a value of each index, which is associated with each index by the "blooming" reflection, to some extent low and high, and the central value of the set of these values, that is, "blooming". and calculates the certainty factor is reflected, E _3. Above D ₃ . Convert each index value obtained in step 1 to the value of each low and high index that is associated with each index by "smear", and set the central value of these values, that is, "smear"
And calculates the certainty factor is, and, F _2. The attribute of the maximum value among the certainty factors that are the "slit shape" reflection, the certainty factors that are the "blooming" reflection, and the certainty factors that are "smear", that is, the "slit shape",
"Blooming" reflections or "smears" are determined as attributes of the peak area. 6. The center of each of the edge range in which the brightness rises and the edge range in which the brightness falls is calculated for the peak area determined as the "slit shape" based on the attribute detection according to claim 1, 4 or 5. A slit center position detecting method in which the calculated center between the centers is determined as the center of the peak area. 7. The center of each of the edge range in which the brightness rises and the edge range in which the brightness falls in the peak area determined as the "slit shape" by the attribute detection according to claim 4 or 5, and the calculated center When there is a scanning line determined to be "blooming" by determining the center of the interval as the center of the peak area, the intensity of the slit light applied to the object is reduced and the image is taken again. The scanning line determined to be "blooming" is subjected to the attribute detection according to claim 4 or 5, and the edge range and the luminance of the peak region of the scanning line determined to be the "slit shape" and the luminance are decreased. A slit center position detecting method in which the center of each edge range is calculated and the center between the calculated centers is determined as the center of the peak area. 8. The center of each of the edge range in which the brightness rises and the edge range in which the brightness falls in the peak area determined as the "slit shape" by the attribute detection of claim 5 is calculated, and between the calculated centers. When the center is determined to be the center of the peak area and there is a scanning line determined to be "blooming", the intensity of the slit light applied to the object is reduced and the image is taken again. The edge range in which the brightness rises and the edge range in which the brightness falls in the peak area of the scan line determined to be the "slit shape" through the attribute detection of claim 4 or 5 for the scan line determined to be "mingling". The respective centers are calculated, the center between the calculated centers is determined as the center of the peak area, and the peak area determined to be "smear" is excluded from the target for determining the center of the peak area. , Slit center position detection method 9. The method for detecting the slit center position according to claim 6, 7 or 8, wherein the calculation of the center of the edge range includes the following steps: a. Calculating a difference in luminance between adjacent pixels in the direction in which the scanning line extends, that is, a differential value, b. The differential value changes from 0 to another value and returns to 0. The differential value of each pixel in a predetermined area within the edge range in the direction in which the scanning line extends is converted into the value of each index such as small, medium, large, c ₁ ． For each of the pixels in the predetermined area, each index value is converted into each edge existing width according to the edge existing width conversion function associated with each index, and a composite center value of each edge existing width is calculated, and , D ₁ . The position of each pixel in the predetermined area in the direction in which the scanning line extends is weighted by the reciprocal of the composite center value, and the center position in the direction in which the scanning line in the predetermined area, which is a set of these pixels, is calculated. 10. The slit center position detecting method according to claim 6, 7 or 8, wherein the calculation of the center of the edge range includes the following steps: a. Calculating a difference in luminance between adjacent pixels in the direction in which the scanning line extends, that is, a differential value, b. The differential value changes from 0 to another value and returns to 0. The differential value of each pixel in a predetermined area within the edge range in the direction in which the scanning line extends is converted into the value of each index such as small, medium and large, c ₂ ． For each of the pixels in the predetermined area, each index value is converted into each edge probability according to the edge probability conversion function associated with each index, and a composite center value of each edge probability is calculated, and d ₂ ． The position of each pixel in the predetermined area in the direction in which the scanning line extends is weighted by the combined center value, and the center position in the direction in which the scanning line in the predetermined area, which is a set of these pixels, is calculated. 11. The slit center position detecting method according to claim 6, 7, or 8, wherein the calculation of the center of the edge range includes the following steps: a. Calculating a difference in luminance between adjacent pixels in the direction in which the scanning line extends, that is, a differential value, b. The differential value changes from 0 to another value and returns to 0. The differential value of each pixel in a predetermined area within the edge range in the direction in which the scanning line extends is converted into the value of each index such as small, medium, large, c ₁ ． For each of the pixels in the predetermined region, each index value is converted into each edge existing width according to the edge existing width conversion function associated with each index, and the first combination that is the combined center value of each edge existing width Calculate the median value, c ₂ . For each of the pixels in the predetermined area, each index value is converted into each edge probability according to the edge probability conversion function associated with each index, and a second combined central value that is a combined central value of each edge probability is calculated. It is calculated, and, d _3. The position in the extending direction of the scanning line of each pixel in the predetermined area is weighted by the first and second combined center values to calculate the center position in the extending direction of the scan line of the predetermined area, which is a set of these pixels. .