JP2005043174A - Color irregularity inspection apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a color irregularity inspection apparatus capable of inspection close to visual inspection. <P>SOLUTION: The display of a display to be inspected 2 is transmitted through a variable filter 21, acquired by a CCD camera 22, and processed by an inspection PC 11. After the inspection PC 11 converts an acquired RGB image into a uniform color space image, a multiplex-resolution image made of a low-resolution image, an intermediate-resolution image, and a high-resolution image is computed in each Luv. An Luv image of each resolution image is approximately differentiated to compute absolute values (space differential images). Each of them is weighted [Mp(x, y)=ä(KlpΔLp(x, y))<SP>2</SP>+(KupΔUp(x, y))<SP>2</SP>+(KvpΔVp(x, y))<SP>2</SP>}<SP>1/2</SP>]. A color difference ((Ip)) of each resolution (p) is weighted (W(p)) for each region to acquire a maximum value, and a maximum value is acquired from the maximum value of every region and evaluated as a color irregularity value. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a color unevenness inspection apparatus that inspects color unevenness and the like generated in a display device such as a plasma display or a liquid crystal display.
[0002]
[Prior art]
Conventionally, as a method of inspecting a display device such as a plasma display or a liquid crystal display, detection is performed by the number or area deviating from the upper and lower limits of the RGB luminance value averaged in each pixel or partial region of an image obtained by photographing the inspection object. Have been devised, and devices that detect the difference between the brightness values of adjacent areas.
[0003]
[Problems to be solved by the invention]
Here, since human beings comprehensively judge information on brightness, color and unevenness edge portion change rate, etc., if the luminance difference is the same, even if the area is large, color unevenness with a small edge change rate is allowed. On the other hand, it is necessary to inspect the color unevenness with a small area and a large edge portion change rate as NG.
[0004]
However, in the above-described conventional inspection apparatus, since the determination is made based on the threshold value of the luminance difference, if the luminance difference is the same, the color unevenness of the area larger than the color unevenness of the small area is judged as NG, which is consistent with the visual inspection. We were unable to perform the inspection.
[0005]
The present invention has been made in view of such a conventional problem, and an object thereof is to provide a color unevenness inspection apparatus capable of performing an inspection close to a visual inspection.
[0006]
[Means for Solving the Problems]
In order to solve the above-described problem, the color unevenness inspection apparatus according to the present invention includes a color unevenness inspection apparatus for inspecting a display device, an input unit for inputting RGB images from the display screen of the display device, and the input Converting means for converting the RGB image input by the means into L, u, and v uniform color space images representing lightness, and multi-resolution images having different resolutions P for the L, u, v components of the uniform color space image, respectively. Multi-resolution image generating means for generating Lp, Up, Vp, spatial differential image generating means for generating spatial differential images ΔLp, ΔUp, ΔVp for each resolution P image Lp, Up, Vp, and for each resolution P Different pixels ΔLp (x, y), ΔUp (x, y), ΔVp (x, y) of the images ΔLp, ΔUp, ΔVp of each resolution P by different weighting factors Klp, Kup, Kvp. Te, a calculating means for calculating a Mp (x, y) by the following equation (1),
[0007]
Mp (x, y) = {(KlpΔLp (x, y)) ² + (KupΔUp (x, y)) ² + (KvpΔVp (x, y)) ² } ^1/2 (1)
[0008]
An evaluation unit that obtains the maximum value Ap of Mp (x, y) for each resolution P, obtains the maximum value A of the maximum value Ap for each resolution, and evaluates the obtained maximum value A as an evaluation value of unevenness. I have.
[0009]
That is, an RGB image input by shooting or the like is converted into a Luv uniform color space, information on lightness (L) and color (u, v) is obtained, and each image of the Luv component is converted into a plurality of multi-resolution images. Each resolution image is spatially differentiated, and unevenness is detected by the maximum value of all the numbers obtained by adding different weights to the values of the respective pixels of the spatial differential image for each Luv component and for each resolution.
[0010]
Accordingly, the degree of attention to the change in brightness and color is changed according to the area of the unevenness and the degree of the edge, and the unevenness matching with the visual inspection is detected.
[0011]
Preferably, in the multi-resolution image, the weight coefficient Klp corresponding to the brightness used in the calculation of the high resolution image is set to a value larger than the weight coefficient Klp used in the calculation of the low resolution image.
[0012]
That is, in visual inspection, small unevenness tends to emphasize lightness, and large smooth unevenness tends to emphasize color (small unevenness is detected with a high-resolution image, and large unevenness is detected with a low-resolution image). Therefore, the weight coefficient Klp of the L (brightness) component is set to be relatively large in the high resolution image and relatively small in the low resolution image.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing a color unevenness inspection apparatus 1 according to the present embodiment. This color unevenness inspection apparatus 1 inspects color unevenness and brightness unevenness generated in a display 2 to be inspected such as a plasma display or a liquid crystal display. In addition to the above, it is an apparatus for inspecting dot-like unevenness and linear unevenness.
[0014]
The color unevenness inspection apparatus 1 is configured around an inspection PC 11, and the inspection PC 11 is configured to output a display signal to the display 2 to be inspected. Accordingly, a graphic display such as gray or RGB having a uniform density according to a signal from the inspection PC 11 can be displayed on the display 2 to be inspected. When white is lit and when black is lit, white (gray) is displayed. On the other hand, it is configured so that it can be determined whether there is color unevenness. In addition, the inspection for lighting each of RGB is configured to be able to inspect whether other colors are lit.
[0015]
The color unevenness inspection apparatus 1 includes a variable filter 21 and a CCD camera 22 that passes through the variable filter 21 and captures a display on the display 2 to be inspected. Since this CCD camera 22 employs monochrome and has an analog portion of 1CH, the balance (electrical characteristics) between channels is more stable than that of an RGB3CH camera. In addition, in order to obtain one image, it is possible to improve noise resistance by inputting and averaging a plurality of images under the same conditions.
[0016]
The variable filter 21 includes a color filter 31 including R, G, B, and colorless filters, and a neutral density filter 32 including a dark color filter and a transparent filter. The filter control unit 33 that outputs a control signal is controlled based on a signal from the inspection PC 11, and the filters are selected.
[0017]
That is, for color unevenness inspection during white display, image input is performed while switching the color filter to R, G, and B. In the case of inspection with high luminance display, a dark neutralization filter is selected, and in the case of inspection with low luminance display, a transmission filter is selected and input to the camera. As a result, the color separation and the dynamic range can be expanded.
[0018]
The CCD camera 22 is connected to the inspection PC 11 via the camera interface 41, and an image photographed by the CCD camera 22 is sent to the inspection PC 11.
[0019]
The inspection PC 11 is configured to operate in accordance with a program, and is configured to perform processing corresponding to the following basic idea.
[0020]
(1) Color unevenness and luminance unevenness occur when the chromaticity and luminance of a certain region (region 1) differ from the chromaticity and luminance of another region (region 2).
[0021]
(2) The inspection is performed by comparing the chromaticity difference and luminance difference between the region 1 and the region 2 with a predetermined threshold value.
[0022]
(3) This threshold value is changed according to the distance between the region 1 and the region 2. That is, when the distance between the region 1 and the region 2 is large, the threshold value is increased (inspection is unsatisfactory), and conversely, when the distance is close, the threshold value is decreased (inspection is strict).
[0023]
(4) If the composition itself is near the center of the display, it will be strict, and if it is in the corner, it will be inspected sweetly.
[0024]
(5) If the above inspection is performed for each pixel of the image, the number of combinations becomes enormous and it is easy to be influenced by camera noise.
[0025]
That is, in the color unevenness inspection algorithm, as an image input process, after selecting an appropriate filter, the CCD camera 22 inputs an image a plurality of times and averages it. In the multi-resolution image processing, as shown in FIG. 2, spatial averaging is performed from the averaged original image 51, and a low resolution image 52, a medium resolution image 53, and a high resolution image 54 having different resolutions are obtained. A multi-resolution image consisting of
[0026]
In the multi-resolution image processing of the present embodiment, the case where three resolution images of the low resolution image 52, the medium resolution image 53, and the high resolution image 54 are generated has been described as an example. However, the present invention is not limited to this. Needless to say, it is better to generate a larger number of images.
[0027]
At this time, in the low-resolution image 52 that is a coarse image, overall unevenness is expressed with a small amount of information, but local unevenness is eliminated. Further, the high-resolution image 54 that is a fine image has a feature that local unevenness is maintained.
[0028]
In the color unevenness inspection process, RGB three-channel images are input from the CCD camera 22 and are converted pixel by pixel into a uniform color space. Specifically, after converting the output value of the RGB filter or the measured value when RGB is individually displayed into the XYZ color system, a generally well-known uniform color space (CIELAB or CIEUV) ). Then, a relative color difference or an absolute color difference between the two points is obtained in the uniform color space.
[0029]
The color difference between the two colors is defined as (ΔL, Δa, Δb) where the difference between the chromaticity values is (ΔL, Δa, Δb).
[0030]
{(ΔL (x, y)) ² + (Δa (x, y)) ² + (Δb (x, y)) ² } ^1/2
Or
[0031]
{(ΔL (x, y)) ² + (Δu (x, y)) ² + (Δv (x, y)) ² } ^1/2
Actually, weight each one,
[0032]
Mp (x, y) = {(KlΔL (x, y)) ² + (KuΔu (x, y)) ² + (KvΔv (x, y)) ² } ^1/2
Is the color difference.
[0033]
At this time, Kl, Ku, and Kv are weight constants, and each weight is determined by surface unevenness (color difference at a distant distance), unevenness of edges (color difference at a close distance), unevenness of a line or a point. Change.
[0034]
That is, as shown in FIG. 3, in the case of an original image in which gradation is applied vertically and the white portion 61 is present in the center, the low resolution image 52 has a large difference between the adjacent pixels in the upper and lower sides, and the medium resolution image. In 53, the difference between the adjacent pixels in the vertical direction becomes small. In the high-resolution image 54, since a large difference occurs between the white portion 61 present in the middle and the upper portion thereof, the weight constants of Kl, Ku, and Kv are set taking these into consideration. .
[0035]
The implementation of such a color unevenness inspection algorithm will be described in accordance with the processing procedure of the program stored in the recording area of the inspection PC 11 as shown in FIG. (High brightness) display is performed (S1), and after the neutral density filter is selected (S2), each of the RGB filters is sequentially selected and each image is input multiple times, and each is averaged to generate an RGB image (S3 to S8), unevenness inspection (in this case, unevenness inspection) is performed (S9).
[0036]
It should be noted that the same processing as described above is performed in 72 when black (low luminance) is displayed on the inspection display and 73 to 75 when each color of RGB is displayed on the inspection display. At this time, in each of the inspection steps 73 to 75 in which only the RGB colors are lit, a filter corresponding to the lit color and a neutral density filter are combined, and the other color filters are used in combination with a transmission filter.
[0037]
In this unevenness inspection, after each RGB (three-dimensional) image is converted into information of a generally well-known uniform color space image (L, U, V), a low resolution image 52 A multi-resolution image (Lp, Up, Vp) composed of the medium resolution image 53 and the high resolution image 54 is calculated.
[0038]
When the gradation unevenness inspection is performed, neighborhood differentiation is performed on each of the LUV images of the resolution images 52 to 54 (mainly the low resolution image 52) to obtain an absolute value (spatial differential image). By this processing, it is possible to grasp how much the color difference between the target pixel and the vicinity is.
[0039]
More specifically, for example, the differential image for L (x, y) is
ΔL (x, y) = El (x, y) ² + E2 (x, y) ² } ^1/2
[0040]
E1 (x, y) = L (x−1, y−1) + 2L (x, y−1) + L (x + 1, y−1) −1 L (x−1, y + 1) −2L (x, y + 1) − L (x + 1, y + 1)
[0041]
E2 (x, y) = L (x-1, y-1) + 2L (x-1, y) + L (x-1, y + 1) -L (x + 1, y-1) -2L (x + 1, y)- L (x + 1, y + 1)
[0042]
L represents density, and x and y represent pixel coordinates.
[0043]
Next, a weighted color difference is calculated. Since the “difference” is obtained by the above differentiation, a weight (K1, Ku, Kv) is added, and the square sum of the differential values of the LUVs may be taken (the square root after the sum). At this stage, 3 channels of LUV become color difference information of 1 channel.
[0044]
This is expressed by the following equation (1). P represents each resolution.
[0045]
Mp (x, y) = {(KlpΔLp (x, y)) ² + (KupΔUp (x, y)) ² + (KvpΔVp (x, y)) ² } ^1/2 (1)
[0046]
Then, the resolution image 52 to 54 is divided into a large number of screens (for example, 7 × 4 divisions), and the color difference information is averaged in each region. At this time, take the root mean square, average only pixels above a certain value, average pixels more than about 1/2 of the maximum value in the area, or output without averaging the maximum value in the area To do. These output values are one value for each area of each resolution image 52-54.
[0047]
When the weights Klp, Kup, and Kvp at this time are constant regardless of the resolution, the features shown in FIGS. 5 to 7 are extracted.
[0048]
That is, FIG. 5 is a diagram showing characteristics of the high-resolution image 54. FIG. 5A shows a large unevenness 81 indicating an overall unevenness in the original image and a medium unevenness 82 indicating a moderate unevenness. A curve showing a small unevenness 83 indicating a small unevenness is shown. FIG. 5B shows the absolute value of the spatial differential value of the image, and shows curves indicating the large unevenness 84, the intermediate unevenness 85, and the small unevenness 86 after the calculation.
[0049]
FIG. 6 is a diagram showing characteristics of the medium resolution image 53. FIG. 6A shows a large unevenness 91 indicating overall unevenness in the original image and a medium unevenness 92 indicating intermediate unevenness. A curve showing a small unevenness 93 indicating a small unevenness is shown. FIG. 4B shows the absolute value of the spatial differential value of this image, and shows curves indicating large unevenness 94, medium unevenness 95, and small unevenness 96 after the calculation.
[0050]
Further, FIG. 7 is a diagram showing the characteristics of the low-resolution image 52. FIG. 7A shows a large unevenness 101 indicating overall unevenness in the original image and a medium unevenness 102 indicating intermediate unevenness. A curve indicating a small unevenness 103 indicating a small unevenness is shown. FIG. 7B shows the absolute value of the spatial differential value of this image, and shows curves indicating the large unevenness 104, the medium unevenness 105, and the small unevenness 106 after the calculation.
[0051]
In addition, each unevenness 81-83 of FIG. 5, each unevenness 91-93 of FIG. 6, and each unevenness 101-103 of FIG.
[0052]
As can be seen from these, for large unevenness in the original image, the edges become steep in the order of unevenness 84 → 94 → 104. As a result, Mp is maximized in the low-resolution image 104 (see FIG. 7B). Further, when the original image has medium unevenness, Mp is maximized in the medium resolution image 95 (see FIG. 6B). Furthermore, in the case of small unevenness in the original image, Mp is maximized in the high resolution image 86 (see FIG. 5B).
[0053]
Considering these features, the weighting coefficient of the entire Luv is set to be large in the high resolution image 54 and small in the low resolution image 52, so that the unevenness detected at each resolution 86, 95, 104 is at the same level. It becomes possible to evaluate.
[0054]
Further, according to visual evaluation, small unevenness with sharp edges tends to be more conspicuous than luminance unevenness. Therefore, the lightness weighting factor Klp is set larger in the high-resolution image 54 and smaller in the low-resolution image 52 than the Kup and Kvp. The image 52 is set so that an inspection similar to a human visual inspection can be performed with emphasis on the color difference.
[0055]
Specifically, the ratio of each weighting factor is set to about Klp: Kup: Kvp = 10: 2: 1.5 in the high-resolution image 54, and Klp: Kup: Kvp = 2: 1 in the low-resolution image. : Set to about 1. For medium-resolution images, it is desirable to set Klp: Kup: Kvp = 5: 1.8: 1.2.
[0056]
Further, when inspecting the edge unevenness, the LUV of each of the resolution images 52 to 54 (mainly the high resolution image 54) is smoothed and then thinned out. Thereby, noise can be reduced while maintaining the edge. Thereafter, neighborhood differentiation is taken for each image to obtain an absolute value, and the subsequent processing is exactly the same as the processing described above.
[0057]
When evaluating (inspecting) the output in each inspection, the maximum value Ap of Mp for each resolution (p) is taken for each resolution and for each region. Next, since the maximum value Ap is output for each region, the maximum value A is further taken as a color unevenness value. Also, the weighting at the center of the screen is set to be large and the weighting at the periphery is set to be small.
[0058]
If the degree of unevenness is greater than or equal to a certain threshold value, the display to be inspected is determined to be NG.
[0059]
In the present embodiment having the above configuration, the weighting in the low resolution image 52 can be reduced and the weighting in the high resolution image 54 can be increased. For this reason, although it is not conspicuous in the visual inspection, it is possible to make a determination closer to the visual inspection even for a large and gentle unevenness that has a large value in the spatial differential image in the low resolution image 52.
[0060]
Also, the weighting coefficient Klp corresponding to the brightness used in the calculation of the high resolution image 54 is changed by the weighting coefficient Klp used in the calculation of the low resolution image 52 by changing the degree of attention to the change in brightness and color according to the unevenness area and the degree of edge. By setting a value larger than Klp, it is possible to inspect the sharpness unevenness of the edge portion by focusing on the lightness difference rather than the color difference, and the slow unevenness of the edge portion by focusing on the color difference. . Therefore, it is possible to determine the quality of image quality that matches the visual inspection performed by a human.
[0061]
Then, by changing the Luv component and the weight coefficients Klp, Kup, and Kvp for each of the resolutions 52 to 54, sharp small unevenness (weight at high resolution) and slow large unevenness (weight at low resolution) at the edge portion. (In general, the weight of the high-resolution image 54 is larger than the weight of the low-resolution image 52), or unevenness with different brightness and unevenness with different colors It is possible to easily change which one is to be inspected.
[0062]
Furthermore, since the maximum value of the inspection result in each of the resolution images 52 to 54 is selected, it is possible to perform an inspection corresponding to the change in the size of the unevenness.
[0063]
In addition, it is possible to make a determination closer to the visual inspection by changing the weighting for each region, specifically, by increasing the weighting at the center of the screen and decreasing the weighting at the peripheral part.
[0064]
【The invention's effect】
As described above, in the color unevenness inspection apparatus of the present invention, the weighting in the low resolution image can be reduced and the weighting in the high resolution image can be increased. For this reason, although it is not conspicuous in the visual inspection, it is possible to make a determination closer to the visual inspection even for a large and gentle unevenness that has a large value in the spatial differential image in the low resolution image.
[0065]
Also, by changing the degree of attention to the change in brightness and color depending on the degree of edge, emphasis is placed on the difference in brightness for steep unevenness at the edge, and emphasis on the difference in color for slow unevenness at the edge. Can be inspected. Therefore, it is possible to determine the quality of image quality that matches the visual inspection performed by a human.
[0066]
If the Luv component and the weighting factor of each resolution are changed, the inspection is made with emphasis on the steep or slow unevenness of the edge portion, or the unevenness of different brightness or the unevenness of color. It is possible to easily change whether to inspect with emphasis.
[0067]
In addition, it is possible to make a determination closer to the visual inspection by changing the weighting for each region, specifically, by increasing the weighting at the center of the screen and decreasing the weighting at the peripheral part.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an embodiment of the present invention.
FIG. 2 is an explanatory diagram showing a multi-resolution image according to the embodiment;
FIG. 3 is an explanatory diagram showing a change in a value appearing in each resolution image according to the embodiment;
FIG. 4 is a flowchart showing a processing procedure in the embodiment;
FIG. 5 is an explanatory diagram illustrating characteristics of the high-resolution image according to the embodiment;
FIG. 6 is an explanatory diagram illustrating characteristics of the medium resolution image according to the embodiment;
FIG. 7 is an explanatory diagram showing characteristics of the low resolution image according to the embodiment;
[Explanation of symbols]
1 Color unevenness inspection device 2 Display to be inspected 11 Inspection PC
22 CCD camera 52 Low resolution image 53 Medium resolution image 54 High resolution image

Claims (2)

In the uneven color inspection device for inspecting the display device,
Input means for inputting RGB images from the display screen of the display device;
Conversion means for converting the RGB image input by the input means into L, u and v uniform color space images representing lightness;
Multi-resolution image generation means for generating multi-resolution images Lp, Up, Vp of different resolutions P for each of the L, u, v components of the uniform color space image;
Spatial differential image generation means for generating spatial differential images ΔLp, ΔUp, ΔVp of the images Lp, Up, Vp of each resolution P;
With respect to each of the pixels ΔLp (x, y), ΔUp (x, y), ΔVp (x, y) of the images ΔLp, ΔUp, ΔVp of each resolution P by different weighting factors Klp, Kup, Kvp for each resolution P, An arithmetic means for calculating Mp (x, y) by the following equation (1):
Mp (x, y) = {(KlpΔLp (x, y)) ² + (KupΔUp (x, y)) ² + (KvpΔVp (x, y)) ² } ^1/2 (1)
An evaluation unit that obtains the maximum value Ap of Mp (x, y) for each resolution P, obtains the maximum value A of the maximum value Ap for each resolution, and evaluates the obtained maximum value A as an evaluation value of unevenness. Color unevenness inspection device characterized by comprising. The weight coefficient Klp corresponding to the lightness used in the calculation of the high resolution image in the multi-resolution image is set to a value larger than the weight coefficient Klp used in the calculation of the low resolution image. Color unevenness inspection device.

Images