JP3327992B2 - Color misregistration correction device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention generates color shift correction data based on a color distribution generated by a color distribution creating means.
The present invention relates to a color shift correction device that performs color shift correction.

[0002]

2. Description of the Related Art A plane-sequential electronic endoscope captures R, G, and B primary color images one by one every 1/60 second. Therefore, when an image of a moving object is taken, R, R
Since the three primary colors of G and B cannot be captured, the colors of the object are separated. This separated state is called color shift.

A method of correcting color misregistration according to Japanese Patent Publication No. 52-9095 is to compare an original signal with a three-field delayed signal, that is, each primary color signal of the original signal and a primary color signal of the same color three fields before and compare the difference. When the signal exceeds a certain value, it is determined that the position is the moving position of the image where the color breakup occurs, and the gain control and the band control of the carrier signal are performed by the movement position determination signal (pulse), and the carrier color signal is converted. Eliminating or suppressing completely at the moving position of the image to reduce color breakup of the three primary colors.

Further, there is disclosed a field sequential color television imaging system for reducing color breakup, in which two or more switching systems for switching to a narrower band pass filter when motion is large may be used.

[0005]

Problems to be Solved by the Invention Japanese Patent Publication No. 52-9095
In this case, since the band of the color signal is limited in accordance with the color shift, the color of the portion with larger motion becomes weaker and becomes closer to a monochrome image, resulting in an unnatural image different from the original subject color.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and has been made in consideration of the above-described problem. The purpose is to provide.

[0007]

In order to achieve the above object, a color shift correcting apparatus according to the present invention compares an image signal having a time difference from a color image signal obtained by an image pickup means for picking up an object in a frame sequential manner. a change section calculating means for calculating a change point or change the region of the image corresponding to the color shift Te, before
The object selected based on the calculation result of the change part calculating means;
A color distribution creating means for creating a color distribution based on a color image signal having no body color shift; and a color space distribution state of a specific color among the color distributions created by the color distribution creating means.
Comprising a color shift correction data selecting means for color shift correction data for estimating color shift corrected color corresponding to a color, and a corrected image generating means for correcting the color shift of the image using the color shift correction data Tei Ru.

[0008]

Since the endoscope image is different from the normal image in the body cavity, the colors constituting the image are considerably deviated, so that the pixel density of a part of the color space is increased. Here, if a representative color having a high pixel density is searched in the color space, the color shift image can be corrected with a color close to the inside of the body cavity.

That is, from the color space distribution state of a particular color in the color distribution created by the color distribution creating means,
By estimating the corresponding color and using this as color shift correction data for color shift correction to correct the color shift of the image, a natural corrected image without a sense of incongruity can be obtained.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. FIGS. 1 to 8 relate to one embodiment of the present invention.
FIG. 2 shows a basic configuration of a color shift correcting device according to one embodiment, FIG. 2 shows a configuration of a color change detecting unit, FIG. 3 shows a configuration of a corrected image creating unit, and FIG. FIG. 5 shows a flow of creating a histogram, FIG. 6 shows an explanatory diagram for obtaining a representative color having a maximum value of the number of pixels from a color space histogram, and FIGS. FIG. 8 shows an explanatory diagram for obtaining a look-up table of a representative color.

As shown in FIG. 1, a color shift correcting apparatus 10 according to a first embodiment includes a color change detecting section 1 for detecting a color change pixel such as a color shift from a signal of an input original image, The image processing apparatus includes a correction image generation unit 2 that generates a correction image of a color change portion, and a synthesis unit 3 that synthesizes the original image and the correction image based on the color change detection unit 1.

As shown in FIG. 2, the color change detecting section 1 receives a luminance signal Y and two color signals C from input R, G and B signals.
matrix circuit 11 for calculating r and Cb, four-field delay memories 12a and 12b for respectively delaying the calculated color signals Cr and Cb signals by four fields, and matrix circuit 11 for color signals Cr and Cb delayed by four fields
Difference detection circuits 13a and 13b for detecting the difference between the color signals Cr and Cb calculated by
An OR circuit 14 for calculating the OR of the detection result of b.

The change point detection signal output from the OR circuit 14 is input to the corrected image forming unit 2 and used for selecting a pixel without color shift, and is also input to the synthesizing unit 3 to output the original image and the corrected image. Used for switching.

As shown in FIG. 3, based on the result of the color change detecting unit 1, the corrected image creating unit 2
G, B signals), a histogram creating circuit 21 for creating a histogram for one field, a CPU block 22 for creating a representative color from the histogram data of the histogram creating circuit 21, and a color for creating a correction color from an original image. The conversion unit 23 includes a conversion unit 23 and a calculation unit 24 that performs a matrix operation on the luminance signal Y of the original image and the correction color of the color conversion unit 23 to create a correction image. The calculation unit 24 outputs the correction image.

The histogram creating circuit 21 is a selector 21 for selecting the input R, G, B signals and the signal read from the last image field memory 212.
1 and a last image field memory 212 for storing the signal selected by the selector 211 for one field.
And a histogram LSI 213 for creating a color distribution from signals for one field stored in the last image field memory 212.

The CPU block 22 includes a histogram LSI.
A representative color is selected from the histogram data created in 213, and a look-up table is created.

The color conversion means 23 receives the input R, G, B
Field memory 2 for delaying signals by one field time
31a, 231b, 231c and a look-up table 23 for recording the representative colors selected by the CPU block 22
2a to 232f, multipliers 233a to 233f for weighting the converted colors, and adders 234a and 23 for adding the color signals weighted by the multipliers 233a to 233f.
4b.

The operation unit 24 includes the matrix circuit 11
A field memory 241 for delaying the luminance signal Y generated by the above by one field time, and adders 234a and 234b
A matrix circuit 24 for creating an image from the color signal created in step (1) and the luminance signal Y delayed in the field memory 241
The color signals R ′, G ′, and B ′ of the corrected image output from the calculation unit 24 are input to the synthesis unit 3.

The synthesizing unit 3 switches between the original image and the corrected image generated by the calculation by the matrix circuit 242 based on the color shift detection result of the color change detecting unit 1, and detects the color shift even if the color shift exists. Output fewer images. This synthesis unit 3
Is composed of, for example, an analog switch.

FIG. 4 shows an electronic endoscope apparatus 71 provided with the color misregistration correction apparatus 10 according to one embodiment.

The electronic endoscope device 71 includes an electronic endoscope 72.
And a light source unit 73 for supplying illumination light to the electronic endoscope 72.
A video processor 75 having a built-in signal processing unit 74 for performing signal processing on image pickup means of the electronic endoscope 72;
And a color monitor 76 for displaying a video signal output from the video processor 75.

The electronic endoscope 72 has a slender, for example, flexible insertion portion 77, and a large-diameter operation portion 78 is connected to the rear end of the insertion portion 77. A flexible universal cable 79 extends laterally from the vicinity of the rear end of the operation unit 78, and a connector 80 is provided at the tip of the cable 79.

The connector 80 is connected to a video processor 75
To the electronic endoscope 72 from the light source 73.
Are sequentially supplied to the light guide 81. That is, the white illumination light of the lamp 82 passes through the R, G, and B color transmission filters 84R, 84G, and 84B provided on the rotary filter 84 that is driven to rotate by the motor 83, so that the R, G, and B fields are sequentially illuminated. Become light, light guide 8
The light is emitted forward from the front end surface of the light emitting device 1.

A subject, such as an affected part, illuminated by the field-sequential illumination light is imaged on a CCD 86 disposed on the focal plane by an objective lens 85 at the tip, and photoelectrically converted. This CCD8
6 is read out by applying a drive signal from a driver 87 to the amplifier 6, the image signal is read out, amplified by an amplifier 88, converted to a digital signal by an A / D converter 89, and further converted to a R signal through a selector 91. , G, and B memories 92R, 92G, and 92B.

The above R, G, B memories 92R, 92G, 9
The image data sequentially stored in 2B is simultaneously read out,
After being input to the color misregistration correction device 10 of the first embodiment and being subjected to color misregistration correction, it is converted into an analog color signal by a D / A converter section 94, and is input to a color monitor 76 together with a synchronization signal (not shown), thereby obtaining a subject image. Is displayed in color.

Next, the operation of the color misregistration correcting apparatus 10 according to the first embodiment will be described. The color change detector 1 converts the digital R, G, B signals into color signals Cr, Cb in the matrix circuit 11 in order to compare color signals having a time difference of four fields. Note that a color signal normally uses RY and BY, but if an operation is performed using 8 bits for each of R, G and B, the color signal becomes 9 bits.
The color signal obtained by reducing this to an 8-bit width is Cr,
It is indicated by Cb.

The two color signals Cr and Cb are input to four-field delay memories 12a and 12b and are delayed by four field times. In the difference detection circuits 13a and 13b, the color signals Cr and
Cb and the color signals Cr and Cb delayed by the four-field delay memories 12a and 12b are respectively subtracted. If the result is positive, the positive comparison is performed, and if the result is negative, the negative comparison is performed.

A change in color is detected by comparing each with a predetermined value. Note the color signal Cr, so either because they are composed of two kinds of Cb is that there was a color change if put out the determination of the discrepancies two difference detection circuit 13a, 13
The outputs of b are combined by an OR circuit 14 to detect a pixel having a color change, and this OR circuit 14 outputs a change point detection signal when detecting a pixel having a color change. This change point detection signal is output to the corrected image creating unit 2 (histogram creating circuit 21) and the synthesizing unit 3.

In order to create a histogram of a color space from pixel data of an image of one field without color shift, a histogram creating circuit 21 first uses a selector 211 to store an original image and an image of one field. The image of the field memory 212 is selected by the change point detection signal, and the selected pixel having no color change is stored in the last image field memory 212.

The histogram LSI 213 creates a color space histogram by using the upper 3 bits of each of the 8 bits of R, G, and B of the image data for one field stored in the last image field memory 212. In addition,
The field memory 213 storing the pixel data of the image for one field without color shift has no contents when read out due to, for example, a FIFO structure. Therefore, the field memory 213 feeds back to the selector 211 and outputs the pixel without color shift immediately before the color shift occurs. Is stored.

In the CPU block 22, R, G, B = 8
Work to select a representative color from a × 8 × 8 color space. Normal,
When a representative color is selected, the color space is often connected by one line, but this representative color calculation method passes through three lines in the color space in order to select two colors from one color.

The reason for this is that the color of the part where the color misregistration occurs lacks any of R, G, and B, and in the worst case, an image may be formed with one color. This is because the two colors must be estimated. Unlike an ordinary image, an endoscopic image is an image in a body cavity, and therefore the colors constituting the image are considerably deviated. Therefore, it is possible to estimate two colors from one color.

This representative color calculation method will be described with reference to the flowchart of FIG. First, the color space histogram is read out as shown in step S1, then the maximum value (the color with the highest frequency) is selected from each color plane as shown in step S2, and the maximum values R, G and B are determined in step S3. The color signals Cr and Cb are calculated as shown in FIG.

Then, as shown in step S4, the color signal C
After interpolating the values of r and Cb into 8-bit gradation, step S
5, look-up tables 232a to 232
Write to f. Lookup tables 232a to 232
f estimates color signals Cr and Cb from one reference color for each pair.

Look-up tables 232a to 232f for estimating color signals Cr and Cb from one reference color
Will be further described with reference to FIGS. When two other colors are estimated from the reference color from the color space, for example, when R is used as a reference, G and B are selected from the color space and R (original color), G '(estimated color), B' ( Estimated color) 1
Become a pair.

The method uses R, G, B as shown in FIG.
Histogram value (pixel count frequency value) at the color space coordinate position of
The color space coordinates of R (gradation of R) in the color space histogram corresponding to are divided into, for example, 8 from 10H to F0H, and eight GB planes are prepared as shown in FIG.
The highest density portion in each plane is selected, and the G and B coordinate values of that portion are recorded as estimated colors.

For example, on the GB plane when R = F0H, if the highest density portion (maximum pixel frequency value) is the portion of the thick frame shown in FIG. Since the positions are D0H and B0H, respectively, this is used as the estimated color.

These are also searched for the other seven GB planes to obtain eight-point data for estimating G from R and eight-point data for estimating B from R. This is also searched for the RB plane and the RG plane, and a table as shown in FIG. 7 is created.

However, since what is obtained here is the input and output data of R, G, and B and not the color data that is ultimately needed, a matrix operation is performed and the R, Cr, C
8 points are obtained from b points, 8 points of Cr and Cb from G, and 8 points of Cr and Cb from B, and the table shown in FIG. 8 is created. Next, interpolation processing is performed from the 8-point data to 256 points, and the result is stored in a lookup table 232.
Write to a to 232f.

The color conversion section 23 has a look-up table 2 in which the representative colors selected by the CPU block 22 are written.
Multiplier 233a calculates three sets of color data from 32a to 232f.
233f, and added for each color data to obtain color signals Cr and Cb.

Next, the arithmetic unit 24 converts the corrected color signal Cr and Cb generated by the color conversion unit 23 and the luminance signal Y of the original image generated by the matrix circuit 11 into a matrix circuit 242.
To obtain a corrected image of R ', G', B '. Note that the luminance signal Y is input to the matrix circuit 242 after being delayed by the field memory 241 because it takes one field time to create the color distribution.

By switching the corrected image obtained in this way and the original image based on the color change point and outputting the synthesized image, it is possible to correct the color shift without weakening the color. When the color misregistration is detected in this way, by performing correction with the corrected image, it is possible to correct with a natural corrected image without a sense of discomfort.

By setting data for correcting a halation portion in advance in a color space histogram for correcting color misregistration,
Color shift based on halation can be satisfactorily corrected.

Further, an image having no color shift and an original image may be combined in accordance with the amount of color shift. Further, the RY and BY signals having a time difference are compared, and if only one of the image signals changes, only the changed color signal can be replaced with image data in which no color shift occurs. good.

Also, instead of linearly interpolating the data of 8 points for each color calculated from the color space, a one-dimensional LPF
May be performed to perform multiple gradations. When interpolating the data of 8 points for each color calculated from the color space, data of 256 gradations may be obtained by interpolating with a straight line passing through the middle between adjacent points. By interpolating between the data to be processed, it is possible to reduce the flat portion of the estimated data and obtain natural correction data.

In addition, since each correction data is estimated only for the upper bit and the lower bit is output as it is, the original image is output as it is. Even if the 8-point estimated data is used as it is, the correction color changes smoothly because natural correction is performed. Data. The number of divisions of the color space and the color plane is not limited to 8 * 8 * 8 and 16 * 16, but may be another number. Further, the color space and the color plane are other color-matched color planes (uv color planes and the like) and color-matched color spaces (L *,
*, B * or Y, RY, BY, etc. color space).

[0047]

As described above, according to the present invention, since the correction color is selected by grasping the color distribution state, the color is corrected with a color close to the original subject color regardless of whether the movement is large or small. Therefore, even if color misregistration occurs, high-accuracy color misregistration correction without a sense of incongruity becomes possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of a color misregistration correction device according to one embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a color change detection unit.

FIG. 3 is a block diagram illustrating a configuration of a correction image creation unit.

FIG. 4 is a configuration diagram illustrating a configuration of an electronic endoscope apparatus including a color misregistration correction device according to one embodiment.

FIG. 5 is a flowchart for creating a histogram.

FIG. 6 is an explanatory diagram for obtaining a representative color having a maximum value of the pixel count frequency from a color space histogram.

FIG. 7 is an explanatory diagram showing representative colors of R, G, and B.

FIG. 8 shows color signals Cr, C corresponding to representative colors of R, G, B.
Explanatory drawing which shows b.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Color change detection part 2 ... Correction image creation part 3 ... Synthesis part 10 ... Color shift correction device 11 ... Matrix circuit 12a, 12b ... Four field delay memory 13a, 13b ... Difference detection circuit 14 ... Adder 21 ... Histogram creation part 22 CPU block 23 Color conversion unit 24 Operation unit

Claims (1)

(57) [Claims] 1. A changing unit calculating means for comparing a color image signal obtained by an image pickup means for picking up an object in a frame-sequential manner with an image signal having a time difference to calculate a change point or a change area of an image corresponding to a color shift. And the object selected by the calculation result of the changing unit calculating means.
A color distribution creation means for creating, based color distribution in the color image signal without color shift of Utsushitai, specific color of the color distribution created by the color distribution generation means
Color shift correction data selecting means for estimating a color corresponding to this specific color from the color space distribution state as color shift correction data for color shift correction, and a color of an image using the color shift correction data. A color shift correcting apparatus, comprising: a corrected image generating unit that corrects a shift.