KR20150047100A - Method for inspecting transparent body, and apparatus for inspecting transparent body - Google Patents

Abstract

It is an object of the present invention to prevent erroneous detection of defects due to thermal fluctuation defects from a photographed image obtained by a Schlieren method under a high temperature environment and to prevent the occurrence of deformation on the surface of a transparent body due to the presence of bubbles or foreign substances in a plate- There is provided a transparent body inspection method and a transparent body inspection apparatus capable of detecting both a bright region and a dark region around a region corresponding to bubbles or foreign matter from a captured image.
(Solution) The line selector 11 sequentially selects each line from a picked-up image of a plate-shaped or thin-film transparent body photographed by the Schleger method. The differential processing section (12) sequentially calculates the differential values of the tone values of the pixels in one line. When a pixel whose derivative value exceeds the first differential threshold value or which is less than the second differential threshold value is detected, the labeling section 13 supplies the gradation value and the gradation value of the first gradation Value threshold value and the second tone value threshold value.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a transparent body inspection method and a transparent body inspection apparatus,

The present invention relates to a transparent body inspection method and a transparent body inspection apparatus, and more particularly to a transparent body inspection method and a transparent body inspection apparatus for inspecting a plate-shaped or thin-film transparent body under a high temperature environment.

There has been proposed a defect inspection system for determining whether or not there is a defective area on a plate having transparency such as a glass plate (see, for example, Patent Document 1). Hereinafter, the defect inspection system disclosed in Patent Document 1 will be described.

The defect inspection system disclosed in Patent Document 1 includes a first defect inspection apparatus, a second defect inspection apparatus, and a processing apparatus.

The first defect inspection apparatus includes a first light source projected onto a plate material having transparency, a first camera for collecting a bright image by condensing the transmitted light passing through the plate material, a knife edge formed on the front face of the first camera, Shielding portion. When bubbles or foreign substances are present in the inside of the plate and the surface becomes uneven, in the bright field image obtained by the first defect inspection apparatus, the name portion is formed so as to face the region of the arm portion. Also, in the case where a foreign substance adheres to the surface of the plate, the list portion is not formed in the bright field image obtained by the first defect inspection apparatus.

The second defect inspection apparatus includes a second light source for irradiating illumination light on the surface of the plate member, and a second camera for condensing the reflected light reflected from the surface of the plate member and capturing a bright field reflection image. The second camera is formed on the same side as the second light source when viewed from the plate. In the bright field reflection image obtained by the second defect inspection apparatus, the area of the defect becomes the dark portion.

The processing apparatus identifies the shape of the defect from the area of the bright part and the dark part in the bright field image obtained by the first defect inspection apparatus and the area of the dark part in the bright field reflection image obtained by the second defect inspection apparatus, .

Patent Document 2 describes an image inspection apparatus that performs determination of whether an inspection object is good or bad based on an inspection image obtained by photographing an object to be inspected (a circuit pattern formed on the surface of the sensor chip, etc.). The image inspection apparatus disclosed in Patent Document 2 regards a group of connected pixels among the pixels representing the outline of the pattern in the binarized image of the inspection image as one connection component. Then, this image inspection apparatus extracts connected contours by assigning the same label to each pixel of the same connected component. Patent Document 2 discloses that differential image processing and background difference processing are performed on a test image as a pre-processing before obtaining a binarized image. In Patent Document 2, the free-wheeling method is described as a method of pulverizing treatment. Patent Document 2 describes that the difference between the average luminance value of the inspection image and the background image prepared in advance is extracted as the background difference processing.

Japanese Laid-Open Patent Publication No. 2010-48745 (paragraphs 0008, 0010, 0026, 0028, 0031, Figs. 1 and 4, etc.) Japanese Laid-Open Patent Publication No. 2010-91361 (paragraphs 0009, 0010, 0019, FIG. 4, etc.)

A method of arranging a light source and a camera such that light emitted from a light source and passing through an object is incident on a camera and forming an optical path shielding portion on the entire surface of the camera to take an object, as in the first defect inspection apparatus disclosed in Patent Document 1, Called the leng method. In the case of a non-uniform refractive index of an object, the non-uniformity of the refractive index of an object can be expressed by using the fact that the light path after passing through the object changes along the passing point of the object.

12 is a schematic diagram showing an example of a picked-up image of a glass plate by the Schleger method. Fig. 12 shows an example of a photographed image at the point where bubbles exist in the glass plate and deformation occurs on the surface. In the photographed image, the position becomes darker at the position of the low gradation value, and becomes brighter at the position of the high gradation value. The area 91 corresponding to the bubble becomes black. Areas 92 and 94 that are brighter than the background region 96 and regions 93 and 95 that are darker than the background region 96 are generated in the vicinity of the region 91. Further, the region 91 becomes black, which is darker than the regions 93 and 95. The background area 96 (excluding the areas 92 to 95) becomes gray (i.e., dark halftone). In Fig. 12, the regions 92 and 94 are all white, but the brightness in the regions 92 and 94 is not necessarily the same. Likewise, darkness in the regions 93 and 95 can not be said to be the same. In Fig. 12, the regions 91 to 95 are schematically shown. Although the contour of the area 91 can be seen relatively clearly, the contours of the areas 92 to 95 are not necessarily clear. The diffusion directions and areas of the regions 92 to 95 are not limited to the example shown in Fig. The same is true when foreign substances other than bubbles are present in the glass plate and deformation occurs on the surface of the glass plate.

Figs. 13 and 14 show examples of images obtained by binarizing a photographed image shown in Fig. 12. In the case of binarizing the image shown in Fig. 12 obtained by the Schleier method, one threshold value is formed, and the tone value is divided into a region having a threshold value or more and a region having a threshold value or less. Therefore, in the binarized image, as shown in Fig. 13, only the region corresponding to the region 92 or 94 can be shown, or only the region corresponding to the region 93 or 95 can be represented as shown in Fig. 14 have. However, each of the regions 92 to 95 around the region 91 corresponding to bubbles can not be represented in one binarized image. Therefore, the size and shape of defects (bubbles and foreign matter) can not be specified from the binarized image. When a binary image is used, the feature amount of the defect can not be calculated based on each of the areas 92 to 95. [ In the technique described in Patent Document 2, the differential processing and the background differential processing are performed as the preprocessing of the binarization processing, but even when such preprocessing is performed, the same problem occurs in the binarized image.

The inspection of the glass plate is carried out after the melted glass is formed into a glass plate. Therefore, inspection of the glass plate is performed under a high temperature environment. When the photographing of the glass plate by the Schlieren method is performed in a high temperature environment, a thermal fluctuation defect occurs on the photographed image. 15 is a schematic diagram showing an example of a thermal fluctuation defect. The thermal fluctuation defect is a stripe-like appearance that occurs in one direction in the photographed image. The thermal fluctuation defect appears as a stripe 97 (see Fig. 15) on the image due to the temperature difference of the air under a high temperature environment. The thermal fluctuation defects include stripes darker than the background area 96 and stripes brighter than the background area 96. The thermal fluctuation defect appears on the photographed image as bright streaks or dark streaks that are brighter than the background region 96, even if there is no defect in the glass plate itself to be inspected. Therefore, when the defect of the glass plate is detected based on the photographed image obtained by the Schleger method under a high-temperature environment, there arises a problem that erroneous detection increases. That is, there arises a problem that it is often determined that a defect has occurred at a point where the defect has not originally occurred.

When inspecting a transparent substance in a plate or thin film form, not only the inspection of the glass plate, but also the application of the Schleger method under a high temperature environment, it is preferable that the defect can be prevented from being erroneously detected due to a thermal fluctuation defect. When bubbles or foreign matter are present in a plate-like or thin film-like transparent body, when deformation occurs on the surface of the transparent body, both the bright region and the dark region around the region corresponding to bubbles or foreign matter can be detected from the photographed image obtained by the Schleen method .

SUMMARY OF THE INVENTION It is therefore an object of the present invention to prevent erroneous detection of a defect due to thermal fluctuation defects from a photographed image obtained by a Schlieren method under a high temperature environment and to prevent the occurrence of defects on the surface of a transparent body by the presence of bubbles or foreign substances in a plate- And it is an object of the present invention to provide a transparent body inspection method and a transparent body inspection apparatus capable of detecting both a bright region and a dark region around a region corresponding to bubbles or foreign matter from a captured image when deformation occurs.

A method for inspecting a transparent body according to the present invention is a method for inspecting a transparent body (for example, a glass plate) from a light source (for example, a light source 3) , A camera (for example, a line sensor camera 1) on which a shading plate is disposed on the front side, generates a photographed image of a transparent body by photographing a transparent body, sequentially selects each line of the photographed image line by line, And when the differential value exceeds a first differential threshold value in one line or is less than a second differential threshold value defined as a value smaller than the first differential threshold value in one line (For example, a starting pixel) is detected, it is determined whether or not the gradation value exceeds a first gradation value threshold value for a pixel beyond the pixel in one line, Value is equal to or smaller than the first tone value threshold value or less than the second tone value threshold value is set to be equal to or greater than the second tone value threshold value and equal to or less than the first tone value threshold value And a region of the pixel whose tone value exceeds the first tone value threshold value and a region of the pixel whose tone value is below the second tone value threshold value are specified.

In the case of calculating the differential value of the tone value of the pixel, the differential value by the difference method is performed once or more based on the predetermined number of pixels behind the pixel and the predetermined number of pixels ahead, do.

In the case of calculating the differential value of the tone value of the pixel, the differential value of the pixel may be calculated by performing the differential by the difference method twice based on three pixels behind and the three pixels ahead .

In the case of calculating the differential value of the tone value of the pixel, the derivative by the difference method is performed once based on the two pixels behind the pixel and the two pixels ahead, thereby calculating the differential value of the pixel You can.

When a pixel whose differential value exceeds the first differential threshold value or is below the second differential threshold value in one line is detected, the calculation of the differential value with respect to the pixel beyond that pixel within one line is stopped And when a pixel whose tone value is in a range not less than the second tone value threshold value and not more than the first tone value threshold value is detected, the calculation of the differential value of the tone value from the next pixel of the pixel may be resumed.

A region of a pixel whose tone value exceeds the first tone value threshold value and a region of a pixel whose tone value is below the second tone value threshold value may be calculated.

The transparent body may be a glass plate.

The transparent body inspection apparatus according to the present invention is provided with a light source (for example, a light source 3) for irradiating light and a plate-shaped or thin-film transparent body to be inspected, (For example, a line sensor camera 1) in which a shading plate is disposed. The camera generates a photographed image of a transparent body by photographing a transparent body, selects a line for sequentially selecting each line of the photographed image line by line Differential processing means (for example, differential processing section 12) for sequentially calculating the differential values of the tone values of the pixels in the selected one line, (For example, a starting pixel) whose differential value exceeds the first differential threshold value or is less than the second differential threshold value that is determined to be smaller than the first differential threshold value, In line It is determined whether or not the gradation value exceeds the first gradation value threshold value, the second gradation value threshold value is equal to or greater than the first gradation value threshold value, and the second gradation value threshold value is less than the second gradation value threshold value Discrimination means (for example, the labeling section 13) for executing the processing for discriminating whether or not the tone value is within the range of the second tone value threshold value or more and the first tone value threshold value or less, (For example, an inspection section 14) for specifying an area of a pixel whose gradation value exceeds a first gradation value threshold value and an area of a pixel whose gradation value is lower than a second gradation value threshold value .

According to the present invention, it is possible to prevent an erroneous detection of a defect due to a thermal fluctuation defect from a photographed image obtained by a Schlieren method under a high temperature environment, and further, since bubbles and foreign substances are present in a plate- It is possible to detect both the bright region and the dark region around the region corresponding to bubbles or foreign matter from the photographed image.

1 is an explanatory view showing an example of a transparent body inspection apparatus of the present invention.
Fig. 2 is an explanatory diagram showing an example of calculation of a differential value of a tone value of a pixel; Fig.
Fig. 3 is an explanatory diagram showing an example of calculation of a differential value of a tone value of a pixel; Fig.
4 is an explanatory view showing an example of calculation of a differential value of a tone value of a pixel;
5 is an explanatory diagram showing an example of a first tone value threshold value and a second tone value threshold value;
Fig. 6 is an explanatory diagram showing an example of a pixel to be judged by the labeling section 13; Fig.
7 is a flowchart showing an example of the processing progress of the transparent body inspection apparatus of the present invention.
8 is a flowchart showing an example of processing progress of the transparent body inspection apparatus of the present invention.
Fig. 9 is an explanatory diagram showing an example of a circumscribed rectangle including a specific region in step S14; Fig.
10 is a schematic diagram showing a change in the differential value of each pixel in a line.
11 is a flowchart showing an example of the processing progress in the reference form.
12 is a schematic diagram showing an example of a picked-up image of a glass plate by the Schleger method;
Fig. 13 is a schematic diagram showing an example of an image obtained by binarizing a photographed image shown in Fig. 12; Fig.
Fig. 14 is a schematic diagram showing an example of an image obtained by binarizing a photographed image shown in Fig. 12; Fig.
15 is a schematic diagram showing an example of a thermal fluctuation defect.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the transparent body inspection apparatus of the present invention, a plate-like or thin-film transparent body is to be inspected. In the embodiments described below, the case where the inspection object is plate-shaped transparent is explained as an example. More specifically, the case where the inspection object is a glass plate will be described as an example.

1 is an explanatory view showing an example of a transparent body inspection apparatus of the present invention. 1 includes a light source 3, a line sensor camera 1, and a light shielding plate 4. The light source 3, the line sensor camera 1, In addition, the transparent body inspection apparatus is provided with the calculation device 10. The photographed image of the glass plate obtained by the photographing using the Schleger method is input to the arithmetic unit 10. Then, the arithmetic unit 10 tests the glass plate based on the photographed image.

The light source 3 is arranged so that light is incident on the lens 2 of the line sensor camera 1. The line sensor camera 1 is disposed on the side opposite to the light source 3 when viewed from the glass plate 6 to be inspected. A light shield plate 4 on the knife edge is formed on the front surface of the line sensor camera 1. [ Specifically, the shading plate 4 is disposed so as to block a part of the light incident on the lens 2 from light irradiated by the light source 3 between the focal point position 5 of the lens 2 and the lens 2 . Light that has passed through the focus position 5 of the lens 2 is incident on the lens 2. However, since the light shield plate 4 is present, a part of the light that has passed through the focus position 5 is blocked by the light shield plate 4. [

The glass plate 6 to be inspected passes through the focal point 5 of the lens 2 and is transported in one direction. At this time, the glass plate 6 is transported with its principal surface perpendicular to the optical axis of the lens 2. The line sensor camera 1 photographs a point in the glass plate 6 passing through the focal point position 5 and generates a photographed image of the glass plate 6. [ This photographing mode corresponds to a photographing mode using the Schlieren method.

The environment in which the light source 3, the line sensor camera 1, and the light shielding plate 4 are disposed is a high temperature (for example, 50 DEG C or higher) to be.

If defects (bubbles or foreign matter) are present in the glass plate 6 and deformation occurs on the surface of the glass plate 6, the surface of the glass plate 6 is deformed, A difference in contrast occurs in the photographed image of the glass plate 6 obtained by the method (1). Hereinafter, the case where the defects are bubbles will be described as an example, but the following description is also appropriate when the defects are foreign substances.

A photographed image of the glass plate 6 is input from the line sensor camera 1 to the arithmetic unit 10.

The computing device 10 includes a line selection unit 11, a differential processing unit 12, a labeling unit 13, an inspection unit 14, and an image cutting unit 15.

The line selecting section 11 sequentially selects each line (row) of pixels from the photographed image of the glass plate 6 one line at a time. The line selector 11 selects one line at a time from the head line of the shot image, for example. It is assumed that each line in the photographed image is perpendicular to the direction corresponding to the carrying direction of the glass plate in the photographed image.

The differential processing section 12 sequentially calculates the differential values of the tone values of the pixels within one line selected by the line selection section 11. The differential processing section 12 calculates the differential value of the pixel by, for example, performing the differential one or more times based on the predetermined number of pixels behind the pixel to be calculated and the predetermined number of pixels ahead do. The predetermined number is hereinafter referred to as differential width.

2 is an explanatory diagram showing an example of calculation of the differential value of the tone value of the pixel. Here, the case of calculating the differential value of the pixel 35 shown in Fig. 2 will be described as an example. The case where the differential width is two will be described as an example. The differential processing section 12 calculates the differential widths (for two) of the rear side of the pixel 35 from the sum of the tone values of the pixels of the derivative differential (for two) The differential value of the pixel 35 can be calculated by subtracting the sum of the tone values of the pixels. In the example shown in Fig. 2, the differential processing section 12 calculates the differential value "90" of the pixel 35 by performing calculation of (170 + 180) - (130 + 130).

Here, it is assumed that the differential processing section 12 sequentially calculates differential values from the pixels on the left side. It is assumed that the pixel on the left side of the target pixel 35 corresponds to the rear side, and the pixel on the right side corresponds to the front side.

As described above, the differential processing section 12 calculates the differential value by subtracting the sum of the values of the pixels of the differential width at the rear of the target pixel from the sum of the values of the pixels of the differential width in front of the pixel in question You can. Alternatively, the differential processing section 12 may calculate the differential value by subtracting the sum of the pixel values of the differential division in front of the pixel of interest from the sum of the pixel values of the differential division behind the pixel under consideration . As in these methods, a method of calculating the differential value by calculating the difference between the sum of the values of the pixels of the differential division in front of the pixel in question and the sum of the values of the pixels of the differential division in the rear of the pixel in question, . In the following description, the differential value is calculated by subtracting the sum of the values of the pixels of the differential width at the rear of the target pixel from the sum of the values of the pixels of the differential width in front of the pixel to which the differential processing unit 12 is interested Will be described as an example.

In the above example, the calculation of the differential value by the difference method is performed once, but the calculation of the differential value by the difference method may be repeated n times by repeating the same calculation. Fig. 3 is an explanatory view showing an example of calculation of the differential value by the difference method twice. Fig. 3, the case where the differential width is two will be described as an example. The differential processing section 12 performs a differential operation on the pixels 33 and 34 and the pixels 36 and 37 of the differential differential section at the rear of the pixel 35 under the differential operation, . As a result of the first differential, X1, X2, X3, and X4 are obtained as differential values of the pixels 33, 34, 36, and 37 of the differential width of the rear portion and the front portion of the pixel 35 ). The differential processing section 12 calculates the differential value of the pixel 35 by performing the second differential using these values. For example, the differential processing section 12 calculates the differential value of the pixel 35 by calculating (X3 + X4) - (X1 + X2) and determines the value as the differential value of the pixel 35 You can.

The differential width is not limited to 2. Fig. 4 is an explanatory diagram showing an example of calculation of a differential value when the differential width is 3; Fig. In this example, the case of calculating the derivative value of the pixel 35 will be described as an example. In the case where the number of differentiations by the difference method is one, the differentiating section 12 calculates, for example, from the sum of the gradation values of the pixels of the differentiation fraction (three portions) in front of the pixel 35, The differential value of the pixel 35 may be calculated by subtracting the sum of the tone values of the pixels of the differential width (three) of the rear portion of the pixel 35. [ 4 illustrates a case in which the differential value "138" of the pixel 35 is calculated by performing calculation of (170 + 180 + 178) - (130 + 130 + 130).

When the number of times of differentiation by the difference method is two, the differentiating section 12 selects the pixels 32, 33, and 34 of the differential division at the rear of the pixel 35, The first derivative is also applied to each of the pixels 36, 37, and 38 by the difference method. Y2, Y3, Y4, Y5, and Y6 as the differential values of the pixels 32, 33, 34, 36, 37, and 38 of the differential widths of the rearward and frontward of the pixel 35 as a result of the first- (See Fig. 4). The differential processing section 12 calculates the differential value of the pixel 35 by performing the second differential using these values. For example, the differential processing unit 12 calculates the differential value of the pixel 35 by calculating (Y4 + Y5 + Y6) - (Y1 + Y2 + Y3) It may be determined as a differential value.

In each of the calculation examples shown in Figs. 2 to 4, the case of calculating the differential value of the pixel 35 is exemplified. The differential processing section 12 may perform the same calculation even when calculating the differential values of the other pixels.

Further, for example, in the pixel near the end of one line, there is no backward pixel or forward pixel, so that the differential value may not be calculated. The differential processing section 12 does not have to calculate the differential value with respect to such pixels.

The labeling unit 13 compares the differential values of the pixels sequentially calculated in the selected one line with the first differential threshold value and the second differential threshold value. The first differential threshold value and the second differential threshold value are predetermined as a threshold value to be compared with the differential value calculated by the differential processing section 12. The first differential threshold value is a positive value and the second differential threshold value is a negative value. That is, the second differential threshold value is set to a value smaller than the first differential threshold value.

It is assumed that a pixel whose differential value calculated by the differential processing section 12 exceeds the first differential threshold value or a pixel whose differential value is below the second differential threshold value is newly detected. The pixel is referred to as a starting pixel. When the starting point pixel is detected, the labeling section 13 judges whether or not the gradation value exceeds the first gradation value threshold value for the pixel beyond the starting point pixel in the selected one line, the gradation value exceeds the second gradation value threshold value, Value threshold value or less than the second tone value threshold value, and labels the determination result for the pixel. That is, the information of the discrimination result is associated with the pixel. The labeling section 13 executes this processing until a pixel whose tone value is within the range of the second tone value threshold value or more to the first tone value threshold value or less is detected.

The first tone value threshold value and the second tone value threshold value are predetermined with a threshold value to be compared with the tone value of the pixel. 5 is an explanatory diagram showing an example of the first tone value threshold value and the second tone value threshold value. In the present embodiment, when the first tone value threshold value and the second tone value threshold value are determined on the basis of the median value of the tone values that can be taken by the pixels in the shot image obtained by the line sensor camera 1 (see Fig. 1) As an example. However, a value other than the median of the tone value may be used as a reference value for determining the first tone value threshold value and the second tone value threshold value. In this example, it is assumed that the maximum gradation value is 255 and the minimum gradation value is zero. In this case, the median value of the grayscale value that the pixel in the captured image can take is 128. Then, a value obtained by adding a predetermined value ("20" in this example) to the median value is set as the first tone value threshold value. In addition, a value obtained by subtracting a predetermined value from the median value is set as the second tone value threshold value. If the median value is 128 and the predetermined value is 20, then the first tone value threshold is 128 + 20 = 148. In addition, the second tone value threshold is 128 - 20 = 108. The predetermined value may be other than "20 ". In the following description, the case where the first tone value threshold value is 148 and the second tone value threshold value is 108 is taken as an example.

In Fig. 5, an example of a histogram of a pixel is also schematically shown. The pixel whose tone value exceeds the first tone value threshold value can be regarded as a pixel belonging to a region brighter than the background region (for example, regions 92 and 94 shown in Fig. 12). A pixel whose tone value is less than the second tone value threshold value can be regarded as a pixel belonging to an area darker than the background area (for example, areas 91, 93, and 95 shown in FIG. 12). A pixel in which the gray level value is in the range from the second gray level value threshold value to the first gray level value threshold value is regarded as a pixel belonging to the background area (for example, the background area 96 shown in Fig. 12) . Hereinafter, a region brighter than the background region is simply referred to as a bright region. A region darker than the background region may be simply referred to as a dark region.

Therefore, the labeling section 13 may judge whether the pixel belongs to the bright region, the background region, or the dark region. Such labeling by the labeling section 13 can be referred to as tri-valent labeling. However, the labeling section 13 does not perform the process of replacing the tone value with any of the three kinds of representative values, and the tone value of each pixel is not changed Do not.

Fig. 6 is an explanatory view showing an example of a pixel to be judged by the labeling section 13. Fig. The pixels 33 to 40 shown in Fig. 6 represent pixels in one selected line. The differential values X1 and X2 of the pixels 33 and 34 shown in Fig. 6 are in the range of the second differential threshold value to the first differential threshold value and the pixels 33 and 34 do not correspond to the starting pixel do. In this case, the labeling section 13 does not compare the gradation value of the pixel with the first gradation value threshold value and the second gradation value threshold value with respect to the pixels 33 and 34.

For example, it is assumed that the differential value 90 of the pixel 35 shown in Fig. 6 exceeds the first differential threshold value. Then, the labeling section 13 judges that the pixel 35 is the starting point pixel and judges whether or not the gradation value exceeds the first gradation value threshold value, the second gradation value threshold value or more It is determined whether or not it falls within the range of the first grayscale value threshold value or less, or less than the second grayscale value threshold value, and the determination result is labeled for the pixel. The labeling section 13 executes this processing until a pixel whose tone value is within a range of the second tone value threshold value or more and the first tone value threshold value or less is detected. In the example shown in Fig. 6, the labeling section 13 determines that the tone value of the pixels 36 and 37 exceeds the first tone value threshold value "148 " Is less than the second tone value threshold value " 108 ", and the tone value of the pixel 39 is within the range of the second tone value threshold value "108" And labels the result of discrimination for each pixel 35-39.

The labeling section 13 determines that the tone value of the pixel 39 is within the range of the second tone value threshold value 108 or more and the first tone value threshold value 148 or less, The determination of the next pixel 40 is stopped. When the starting point pixel is detected again, the labeling section 13 resumes the discrimination process from the starting point pixel, and the labeling section 13 resumes the discrimination process so that the gradation value is equal to or higher than the second gradation value threshold value "108" The discrimination process is carried out up to the pixel in the range below.

When the labeling section 13 detects the starting pixel, the differential processing section 12 stops the calculation of the differential value of the tone value for the next pixel after the starting pixel within the selected one line. The labeling section 13 sets the pixel (the pixel 39 in the example shown in Fig. 6) whose gradation value falls within the range of the second gradation value threshold value 108 or more and the first gradation value threshold value 148 or less When detected, the differential processing section 12 resumes the calculation of the differential value of the tone value from the next pixel.

In the example shown in Fig. 6, the differential processing section 12 sequentially calculates the differential values for the pixels 33, 34, and 35. Fig. Then, when the pixel 35 is detected as the starting point pixel, the differential processing section 12 stops calculating the differential value for the pixels after the pixel 36. [ Then, the differential processing section 12 resumes the calculation of the differential value from the pixel 40.

Based on the discrimination result labeled on the pixel by the labeling section 13, the inspection section 14 determines whether or not the area of the pixel whose gray level value exceeds the first gray level value threshold and the area of the pixel whose gray level value is less than the second gray level value threshold The area of the pixel is specified from the photographed image.

The inspection section 14 may calculate the feature quantity of the defect based on the specific area. An example of calculating the feature amount will be described later.

The image cutting unit 15 cuts out a rectangular area including an area specified by the inspection unit 14 from the image input to the arithmetic unit 10.

The line selecting unit 11, the differential processing unit 12, the labeling unit 13, the checking unit 14 and the image cutting unit 15 are realized by a CPU of a computer operating in accordance with a program, for example. In this case, a CPU of the computer reads and inputs a program and operates as a line selection unit 11, a differential processing unit 12, a labeling unit 13, an inspection unit 14, and an image cutting unit 15 in accordance with the program .

Next, examples of the processing progress of the present invention will be described. Figs. 7 and 8 are flowcharts showing examples of processing progress of the transparent body inspection apparatus of the present invention. Fig. The light source 3 irradiates the glass plate 6 conveyed under a high-temperature environment (for example, an environment of 50 ° C or higher). Then, the line sensor camera 1 photographs the glass plate 6 to be conveyed by the Schlieren method. The line sensor camera 1 divides the photographed image of the glass plate 6 photographed by the Schlieren method into a predetermined size (for example, P pixel x Q line size) and inputs the image to the arithmetic unit 10 Step S1).

The line selector 11 of the arithmetic unit 10 selects one line from the shot image input in step S1 (step S2). The line selecting unit 11 selects one line at the head when the process goes to step S2 for the first time.

Next, the differential processing section 12 selects one pixel to be a target for calculating the differential value in the selected one line (step S3). Then, the differential processing section 12 calculates the differential value of the tone value of the selected pixel (step S4). Since an example of the method of calculating the differential value has already been described, the description is omitted here.

Next, the labeling section 13 determines whether the differential value of the pixel being selected (i.e., the differential value calculated in step S4) exceeds the first differential threshold value or is less than the second differential threshold value Step S5).

If the differential value exceeds the first differential threshold value or is less than the second differential threshold value (YES in step S5), the process proceeds to step S6 and subsequent steps with the pixel being selected as the starting pixel. In step S6, the labeling section 13 determines whether the tone value of the pixel being selected exceeds the first tone value threshold value, falls within the range not less than the second tone value threshold value or less than the first tone value threshold value, It is determined whether or not it is less than the second tone value threshold value, and the determination result is labeled for the pixel (step S6). When the tone value exceeds the first tone value threshold value, the labeling section 13 associates the information of the tone value with the pixel. This means that the pixel belongs to the bright region. If the gradation value is less than the second gradation value threshold value, the labeling section 13 associates the information of the gradation value with the pixel. This means that the image belongs to the dark region. When the tone value is within the range of the second tone value threshold value or more and the first tone value threshold value or less, the labeling section 13 associates the information of this effect with the pixel. This means that the image belongs to the background area.

When it is determined that the gradation value of the pixel falls within the range not less than the second gradation value threshold value and the first gradation value threshold value (YES in step S7), the process proceeds to step S12.

If it is determined that the gradation value of the pixel exceeds the first gradation value threshold value or is less than the second gradation value threshold value (NO in step S7), the process proceeds to step S8. Hereinafter, a pixel whose tone value exceeds the first tone value threshold value is referred to as bright pixel. A pixel whose gradation value is less than the second gradation value threshold value is referred to as a dark pixel.

In step S8, the labeling unit 13 specifies a pixel closest to the pixel being selected, among pixels determined to be bright pixels or dark pixels in one line being selected. Here, the type of the bright pixel or the dark pixel of the specified pixel may be different from the bright pixel or dark pixel of the selected pixel. The labeling section 13 determines whether or not the distance between the specified pixel and the pixel being selected is equal to or less than a predetermined threshold value (a threshold value predetermined for determination in step S8). If the distance is less than or equal to the threshold value, the labeling unit 13 assigns the defect ID equal to the defect ID assigned to the specified pixel to the pixel being selected (step S8). The process of step S8 can be referred to as a defect proximity combining process in one line.

After step S8, the labeling unit 13 specifies a pixel closest to the pixel being selected, among pixels judged to be bright pixels or dark pixels, as pixels in other lines that have already been selected. Here, the type of the bright pixel or the dark pixel of the specified pixel may be different from the bright pixel or dark pixel of the selected pixel. The labeling section 13 determines whether or not the distance between the specified pixel and the pixel being selected is equal to or less than a predetermined threshold value (a threshold value predetermined for determination in step S9). If the distance is less than or equal to the threshold value, the labeling unit 13 assigns the defect ID equal to the defect ID assigned to the specified pixel to the pixel being selected (step S9). The process of step S9 can be referred to as a defect proximity combining process with another line.

The defect ID is identification information for identifying a defect. The pixels to which the same defect ID is assigned mean that they are bright pixels or dark pixels due to the same defect (bubble or foreign matter).

In any of steps S8 and S9, if no defect ID is assigned to the pixel being selected, the labeling unit 13 assigns a new defect ID to the pixel being selected.

Subsequently, the labeling section 13 judges whether or not any of the selected lines is not selected in any of steps S3 and S11 (step S10). If there is an unselected pixel (YES in step S10), the next one pixel of the pixel being selected is newly selected from among the lines being selected (step S11). Then, the labeling section 13 executes the processings after step S6 for the pixel. If there is no unselected pixel (NO in step S10), the process proceeds to step S13 (see Fig. 8).

If the differential value of the pixel being selected is within the range of the second differential threshold value to the first differential threshold value (NO in step S5), the labeling section 13 selects one It is determined whether or not any pixel is not selected in any of steps S3 and S11 (step S12). If there is an unselected pixel (YES in step S12), the processing in step S3 and subsequent steps is repeated. In other words, the differential processing unit 12 newly selects one pixel from among the currently selected line (step S3), and performs the processing after step S4 on the pixel. If there is no unselected pixel (NO in step S12), the process proceeds to step S13 (see Fig. 8).

In step S13, the line selector 11 determines whether or not the selection of the last line of the input captured image is completed (step S13). If the last line has not been selected yet (NO in step S13), the processing in step S2 and subsequent steps is repeated. That is, the line selector 11 newly selects the next one line of the line being selected (step S2), and performs the processing after step S3 on the line.

If the final line is selected (YES in step S13), the checking unit 14 specifies a set of pixels to which a common defect ID is assigned. The set of pixels includes a bright pixel (a pixel whose gray level value exceeds the first gray level value threshold) and a dark pixel (a pixel whose gray level value is lower than the second threshold value). The checking unit 14 determines whether or not the area of the set of bright pixels (pixels whose gray level value exceeds the first gray level value threshold) and the dark area of the dark pixels (I.e., pixels less than the threshold value) (step S14). The inspection unit 14 performs the process of step S14 for each defect ID.

After step S14, the inspection unit 14 calculates the feature quantity of the defect on the basis of the area of the set of bright pixels and the area of the dark pixels specified in step S14 for each defect ID. Hereinafter, an example of calculation of the feature amount is shown.

The checking unit 14 compares the total number of pixels of the bright pixels belonging to the set of bright pixels specified in step S14 and the sum of the brightness values of the bright pixels, the number of pixels of dark pixels belonging to the set of dark pixels specified in step S14 , And the sum of the tone values of the dark pixels may be calculated as the feature quantities of the defects, respectively.

The inspection unit 14 may also calculate the number of pixels belonging to the circumscribed rectangle including the region of the set of bright pixels and the region of the set of dark pixels specified in the step S14 as the defect feature amount. The inspection unit 14 may calculate the ratio of the number of pixels belonging to the region of the set of bright pixels and the pixel of the set of dark pixels to the number of pixels belonging to the circumscribed rectangle as the feature amount of the defect. This ratio can be referred to as the density in the circumscribed rectangle. Fig. 9 is an explanatory diagram showing an example of a circumscribed rectangle including the area specified in step S14. It is assumed that the areas 91 to 95 are specified in step S14. The defect IDs assigned to the pixels belonging to the regions 91 to 95 are common. The inspection unit 14 may specify the circumscribed rectangle 99 including the regions 91 to 95 and calculate the number of pixels belonging to the circumscribed rectangle 99 as the feature amount. The inspection unit 14 may calculate the ratio (density) of the number of pixels belonging to the regions 91 to 95 to the number of pixels belonging to the circumscribed rectangle 99 as a feature amount.

Next, the inspection section 14 identifies the type of defect for each defect ID (step S16). More specifically, in the case where only the region of the set of dark pixels is specified in the step S14 with respect to the defect ID of interest, and the region of the set of bright points is not specified with respect to the defect ID of interest, Is judged to be foreign matter adhering to the surface of the glass plate. On the other hand, in the case where the area of the set of dark pixels and the area of the set of bright pixels are specified in step S14 with respect to the defect ID of interest, the defect corresponding to the defect ID is detected by the bubble Or foreign matter.

Next, the image cutting unit 15 cuts out an area including each area, which is a set of pixels assigned a common defect ID for each defect ID, from the input captured image (step S17). The region to be cut in step S17 is an area wider than the circumscribed rectangle of each area which is a set of pixels to which the defect ID is assigned. The image cutting unit 15 displays, for example, the area cut in step S17 on a display device (not shown) formed in the computing device 10. [

The arithmetic operation unit 10 executes the processing after step S2 when a newly captured image of a predetermined size is input from the line sensor camera 1. [

Effects of the present embodiment will be described.

10 is a schematic diagram showing the change in the differential value of each pixel in the line when the dark region 101 is present in the selected line, rather than the background. The gray level values of the pixels in the background region are the same value. Therefore, as shown in Fig. 10, the differential value of the pixels in the background region does not change so much. Since the gradation values of the pixels in the background region and the region 101 are different from each other, the differential value greatly fluctuates in the pixels near the outer edge of the region 101. [ In the present embodiment, a point at which the differential value greatly fluctuates is detected as the starting point pixel.

Although not shown in Fig. 10, under the high-temperature environment, the stripe pattern 97 (heat fluctuation defect) shown in Fig. 15 occurs. Such a thermal fluctuation defect occurs along each line in the photographed image. Even if a thermal fluctuation defect occurs, the pixels corresponding to the thermal fluctuation defect in the line are continuously arranged, and the gradation values of the pixels become almost the same value. Therefore, the differential value of the pixel corresponding to the thermal fluctuation defect in the line also does not fluctuate so much, and the pixel corresponding to the thermal fluctuation defect is not detected as the starting pixel well.

When the starting point pixel is detected, the labeling section 13 determines whether or not the gradation value exceeds the first gradation value threshold value, the second gradation value threshold value, and the first gradation value threshold value Or less than the second tone value threshold value, and labels the determination result for the pixel. The labeling section 13 executes this processing until a pixel whose tone value is within the range of the second tone value threshold value or more to the first tone value threshold value or less is detected.

Therefore, the pixel corresponding to the thermal fluctuation defect is not processed by the labeling section 13, and the discrimination result by the labeling section 13 is not labeled. Therefore, erroneous detection caused by a thermal fluctuation defect can be prevented.

In the present embodiment, when the tone value exceeds the first tone value threshold value, the fact that the tone value is below the second tone value threshold value is labeled in the pixels after the starting point pixel Lt; / RTI > In other words, both of a bright region and a dark region can be detected from a background image in one captured image. Therefore, unlike the binarized image exemplified in Figs. 13 and 14, it is possible to specify the size and shape of a defect using both a bright region and a dark region rather than a background region. Also, as exemplified in step S15, the feature amount of the defect may be calculated based on both the light region and the dark region that are brighter than the background region.

In the above embodiment, in the case of calculating the differential value, the absolute value of the peak value of the differential value shown in FIG. 10 can be increased and the starting pixel can be detected more easily as the differential width and the differential frequency count are larger. However, if the differential width and the number of differentiation are increased, a processing time is required. Therefore, it is particularly preferable that the differential width is 3 and the number of differentials is 2. Alternatively, the differential width may be set to 2, and the number of differentials may be set to one. However, the values of the differential width and the number of differentiation are not limited to these values, but may be other values.

In the above embodiment, the case where the inspection object is a glass plate has been described as an example. However, the object to be inspected in the present invention is not limited to a plate-shaped transparent body such as a glass plate, but may be a thin-film transparent body (for example, a film).

Next, an example of a reference form in which a differential is not performed is shown as an object to be compared with the present embodiment. 11 is a flowchart showing an example of the process progress of the reference form. The operation of the line sensor camera 1 is the same as that of the above embodiment, and the photographed image of the glass plate 6 photographed by the Schleger method is divided by a predetermined size, and the arithmetic unit (not shown in the drawing, (Step S21).

The computing device selects one line from the input captured image (step S22).

Subsequently, the arithmetic operation unit determines whether the gray level value exceeds the first gray level value threshold, whether the gray level value exceeds the second gray level value threshold or below the first gray level value threshold, It is determined whether or not it is less than the second tone value threshold value, and the discrimination result is labeled in the pixel (step S23).

The arithmetic unit determines whether or not at least one pixel whose gradation value exceeds the first gradation value threshold value or a pixel whose gradation value is below the second gradation value threshold value as a result of step S23 exists in one line (Step S24). If such a pixel does not exist (NO in step S24), the process proceeds to step S27. If such a pixel exists (YES in step S24), the arithmetic and logic unit executes the defect proximity combining processing in one line and the defective proximity combining processing with other lines (steps S25 and S26 , The process proceeds to step S27.

In step S27, the computing device determines whether or not the selection of the last line of the input captured image is completed (step S27). If the last line is not yet selected (NO in step S27), the processing in step S22 and subsequent steps is repeated. When the selection of the last line is completed (YES in step S27), the same processing as in step S14 and subsequent steps of the above-described embodiment is performed. That is, the arithmetic unit specifies the region of the set of bright pixels (the pixel whose gray level value exceeds the first gray level value threshold) and the region of the dark pixel (the pixel whose gray level value is lower than the second threshold value) . Further, the computing device calculates the feature quantity of the defect based on the region of the set of bright pixels and the region of the dark pixels. Further, the computing device identifies the type of defect. Then, the arithmetic unit performs image cutting. These processes are the same as those in steps S14 to S17 in the above embodiment, and are not shown in Fig.

In the reference form, since the processing in step S23 is performed, both a bright region and a dark region can be detected in a single captured image than in the background region.

However, in the reference form, since the gradation value of the pixel is not subjected to the differential processing, the step S23 is performed. Therefore, even in the case of the pixel corresponding to the bright stripe or the dark stripe appearing in the captured image as the thermal fluctuation defect, The labeling is performed in the same manner as the pixel in the bright region or the pixel in the dark region. Therefore, in the reference form, both the bright region and the dark region can be detected as compared with the background region, but it is difficult to realize prevention of erroneous detection caused by thermal fluctuation defects.

Therefore, the above-described embodiment is preferable to the above-mentioned reference embodiment.

The present invention is preferably applied to inspection of a glass plate under a high temperature environment.

1: Line sensor camera
3: Light source
4: Shading plate
6: Glass plate
10:
11: Line selection unit
12:
13: Labeling unit
14:
15:

Claims (8)

A light is irradiated from a light source to a plate-shaped or thin-film transparent body,
A camera disposed on a side opposite to the light source when viewed from the transparent body and having a shield plate disposed on the front side to generate a captured image of the transparent body by photographing the transparent body,
Each line of the captured image is sequentially selected by one line,
Sequentially calculates the differential values of the tone values of the pixels in one selected line,
When a pixel in which the differential value exceeds the first differential threshold value in the one line or is less than the second differential threshold value that is determined to be smaller than the first differential threshold value is detected, It is determined whether or not the gradation value exceeds the first gradation value threshold value, the second gradation value threshold value is equal to or greater than the first gradation value threshold value, and the second gradation value threshold value is less than the second gradation value threshold value Is performed until a pixel whose tone value is within a range of a second tone value threshold value or more and a first tone value threshold value or less is detected,
A region of a pixel whose grayscale value exceeds a first grayscale value threshold value and a region of a pixel whose grayscale value is lower than a second grayscale value threshold value. The method according to claim 1,
When calculating the differential value of the tone value of the pixel, calculating the differential value of the pixel by performing the differential by the difference method one or more times based on the predetermined number of pixels behind the pixel and the predetermined number of pixels ahead A method of transparency inspection. 3. The method according to claim 1 or 2,
When the differential value of the tone value of the pixel is calculated, differential values by the difference method are performed twice based on three pixels behind the pixel and three pixels ahead of the pixel to calculate the differential value of the pixel , Transparency inspection method. 3. The method according to claim 1 or 2,
When calculating the differential value of the tone value of the pixel, the differential value by the difference method is performed once based on the two pixels behind the pixel and the two pixels ahead, thereby calculating the differential value of the pixel , Transparency inspection method. 5. The method according to any one of claims 1 to 4,
When a pixel whose differential value exceeds the first differential threshold value or is less than the second differential threshold value in one line is detected, the calculation of the derivative value with respect to the pixels subsequent to the pixel in the one line is performed Stop,
Wherein the calculation of the differential value of the tone value from the next pixel of the pixel is resumed when a pixel in which the tone value falls within the range from the second tone value threshold value to the first tone value threshold value is detected. 6. The method according to any one of claims 1 to 5,
A region of a pixel whose grayscale value exceeds a first grayscale value threshold value and a region of a pixel whose grayscale value is lower than a second grayscale value threshold value. 7. The method according to any one of claims 1 to 6,
Wherein the transparent body is a glass plate. A light source for emitting light,
And a camera disposed on a side opposite to the light source when viewed from a plate-shaped or thin-film transparent object to be inspected, the light-shielding plate being disposed on the front surface,
The camera generates a photographed image of the transparent body by photographing the transparent body,
Line selecting means for sequentially selecting each line of the photographed image line by line;
Differential processing means for sequentially calculating differential values of tone values of pixels in one selected line,
When a pixel in which the differential value exceeds the first differential threshold value in the one line or is less than the second differential threshold value that is determined to be smaller than the first differential threshold value is detected, It is determined whether or not the gradation value exceeds the first gradation value threshold value, the second gradation value threshold value is equal to or greater than the first gradation value threshold value, and the second gradation value threshold value is less than the second gradation value threshold value Discrimination means for discriminating whether or not a process for discriminating a pixel having a gray level value equal to or greater than a second gray level value threshold value and a first gray level value threshold value or less is detected;
And area specifying means for specifying an area of a pixel whose grayscale value exceeds a first grayscale value threshold value and an area of a pixel whose grayscale value is lower than a second grayscale value threshold value.

Images