JPH02116978A - Uniformity evaluating method - Google Patents

Abstract

PURPOSE:To automatically and quantitatively check the uniformity of picture quality or the like by obtaining a block average matrix of a normalized variable density level and obtaining correlations between respective rows or columns of this matrix to evaluate an evaluation object. CONSTITUTION:The white light is radiated from a light box 11, and in this state, an ITV camera 10 picks up the image of this white light, and the image is decoded by a decoder 12 with respect to each of respective colors R, G, and B and is sent to an image processing device 13. The device 13 stores image data of each color in an image memory. The device 13 reads out each image data from the image memory and divides it into N in the row direction and the column direction. The device 13 obtains the average variable density level of each block in accordance with variable density levels of respective picture elements by calculation. The device 13 obtains the block average matrix of the variable density obtained by normalizing the block average variable density level by the average variable density level in the row or column direction, and correlations between respective rows or columns of this block average matrix are obtained to evaluate the evaluation object.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention is applicable to, for example, CCD (solid-state image sensor)
The present invention relates to a uniformity evaluation method suitable for checking image quality defects of liquid crystal display devices (liquid crystal display devices) over a period of one week.

(Prior Art) For example, in COD, image quality defects are inspected at the final stage of the production process. In this inspection, as shown in FIG. 8, the assembled CCD is installed in an ITV (industrial television) camera 1, and a light box 2 is placed in front of the ITV camera 1. And ITV camera 1
The light box 2 is imaged, the image signal is sent to the monitor television 3, and the image is displayed on the monitor television 3. Then, the inspector inspects this image and performs an image quality defect inspection of the CCD, for example, image unevenness (shading inspection).

Specifically, image quality defects are classified into diagonal lines, lines, large areas, and creases depending on their shape, as shown in FIGS. 9(a) to 9(d).

(Problems to be Solved by the Invention) However, since the image quality defect inspection method described above involves a sensory inspection by an inspector, the inspection results may vary depending on the inspector's mental factors. Additionally, inspectors are required to be highly skilled.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a uniformity evaluation method that can be automated and quantitatively inspected for uniformity of image quality, etc.

[Configuration of the Invention (Means and Effects for Solving the Problems) The present invention divides image data to be evaluated into a plurality of blocks in the matrix direction, calculates the average gray level of each block, and calculates the average gray level of each block. A block average matrix of gray level is obtained by normalizing the level by the average gray level in the row or column direction, and then each correlation between rows or columns of this block average matrix is calculated to evaluate the evaluation target. This is a uniformity evaluation method.

Further, the present invention divides the image data to be evaluated into a plurality of blocks in the matrix direction, obtains the average gray level for each block, and normalizes these block average gray levels by the average gray level in the row or column direction. A block average matrix of gray levels is determined, then a matrix representing each correlation between each row or column of this block average matrix is determined, and the uniformity parameter of each component in the diagonal direction of this matrix is determined to be evaluated. This is a uniformity evaluation method that evaluates

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a CCD image quality defect inspection apparatus to which the method of the present invention is applied. In the figure, 10 is an ITV camera incorporating a CCD, and a light box 11 is arranged in front of this ITV camera 10. Further, the image signal output from the ITV camera 10 is sent to the decoder 12, and the decoder 12 converts it into R (red), G (green), and B (
blue) is separately decoded and stored as image data in the image memory within the image processing device 13. This image processing device 13 divides the image data stored in the image memory into a plurality of blocks in the matrix direction, obtains an average gray level for each block, and normalizes these block average gray levels by the average gray level in the row direction. Then, calculate the matrix representing each correlation between each row of this block average matrix, and further calculate each component in the diagonal direction of the matrix representing each correlation. This function performs a function of determining uniformity parameters and evaluating image data by CCD, that is, inspecting image quality defects. Further, 14 is a display device.

Next, a description will be given of CCD image quality defect inspection using the apparatus configured as described above.

While white light is being emitted from the light box 11, the ITV camera 10 images the white light. At this time,
The image signal output from the ITV camera 10 is sent to the decoder 1
2, the signals are decoded separately for each color of RSG and B and sent to the image processing device 13. This image processing device 13
stores image data for each color in an image memory.

Next, the image processing device 13 converts the image data of each color to the CCD.
Perform image quality defect inspection. That is, the image processing device 1
3 reads out, for example, B image data among the R, G, and B image data from the image memory and divides it into a plurality of blocks in the matrix direction as shown in FIG. In this case, the block is divided into N pixels in each of the row and column directions, and the number of pixels in each divided block is m in the row direction and n in the column direction. Note that the positions in the row and column directions in each block are indicated by i and j.

Next, the image processing device 13 calculates each average gray level ak for each block from the gray level X of each pixel by calculating the following equation.

ao-(ΣΣx B)/m-n-(1) Thus, the average gray level ak of each block.

Once obtained, the average gray level μ for each row.

is required. In addition, each average gray level bell μ in this row direction
For example, if m is the first row, the following formula u+ −(a11+aB+−a+N) /N −(
2).

Next, the image processing device 13 calculates the average gray level a of each block.
k, is normalized by the average gray level μ of each row and a′.

shall be. That is, normalization is performed by calculating au-ass u*-(3). In addition, k,,t'-1,2゜...
・N. Then, each position obtained by this normalization is represented by a matrix A shown in the following equation.

ψkJ)−(a−1,,Q−a′1.k)2+°°+(
a'N, 1l-a-N, k)2. By the way, if this evaluation matrix is explained using image data divided into 4x4 blocks as shown in Fig. 3, the first row and the second row
For rows, the square of the difference between the normalized values a′ for each block, that is, (a / , , a / ) 2 ... (a
'14-a'24)2 is obtained. Then, these values are added to form element ψ8. is required. Similarly, for other rows, for example, element ψ1 from the first row and the third row. is required. However, an evaluation matrix W is obtained.

Next, the image processing device 13 uses the evaluation matrix V
Execute matrix operation tlf-A◎A to obtain . Here, the evaluation matrix W is an N×N matrix, and its elements ψ are: This evaluation matrix V represents the mutual correlation in row units of the matrix A, and is the diagonal element ψ in the target matrix.

(k-, i') becomes "0".

As a result, a COD image quality defect inspection is performed from this evaluation matrix W. The present invention may directly evaluate image defects using this evaluation matrix V. For example, this may be done by converting this evaluation matrix W into an image and having the operator judge it.
The determination may be made automatically by a computer.

In this embodiment, next, a parameter P for evaluating image quality defects (image unevenness) is determined. This parameter P indicates the uniformity of components parallel to the axis of symmetry, as shown in FIG. 6, in order to evaluate the stationarity when the image is considered as a stochastic process from the average of the evaluation matrix. In other words, the parameter P
It is calculated by . Here, μ is μg-(, ψ, ni-(I-1))/(N-Ω+1)
Q-2, ..., N).

When the parameter P is determined in this way, the result is the seventh
As shown in the figure, the parameter values are different between a good CCD and a defective CCD.

Furthermore, as can be seen from the figure, the parameter values are different between the oblique image quality defects and the large area and line-shaped image quality defects.

However, the value of the parameter P can also determine the type of image quality defect.

In this way, in the above embodiment, image data is divided into a plurality of blocks in the matrix direction, the block average -a gray level is obtained for each block, and these block average gray levels are normalized by the average gray level in the row direction. As a result, the evaluation matrix W representing each correlation in the row direction of this block average matrix is determined automatically and quantitatively. can. In this case, image quality defects can be more accurately determined by displaying the evaluation matrix W as an image.

In addition, since the uniformity parameter P of each component in the diagonal direction of the evaluation matrix M is determined to determine the image quality defect of the CCD, the image quality defect can be determined as a parameter value. For example, by setting the determination value, the image quality can be determined. Defects can be determined more reliably than by using only the evaluation matrix W, and it is also possible to determine the type of image quality defect.

Note that the present invention is not limited to the above-mentioned embodiments, and may be modified without departing from the spirit thereof.

For example, in the above embodiment, an evaluation matrix indicating the correlation in the row direction is obtained, but an evaluation matrix indicating the correlation in the column direction may be obtained. Furthermore, it can be applied not only to inspecting CCD image quality defects, but also to TFT image quality defect inspection, and furthermore, to evaluating whether or not the coating of refrigerators, automobiles, etc. has been uniformly applied. Furthermore, when inspecting image quality, it can also be applied to the evaluation of periodic images. Furthermore, although evaluation matrices and the like were obtained from digitized image data, evaluation matrices can also be obtained from image signals before digitization. Furthermore, the parameter P obtained from this evaluation matrix can also be modified in various ways to obtain the feature quantity. For example, dividing the above parameter P by the total number of pixels,
The variance may also be calculated.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide a uniformity evaluation method that can be automated and quantitatively inspected for uniformity of image quality, etc.

[Brief explanation of the drawing]

1 to 7 are diagrams for explaining an embodiment of the uniformity evaluation method according to the present invention, in which FIG. 1 is a configuration diagram of an image quality defect inspection apparatus to which this method is applied, and FIG. FIG. 3 is a diagram for explaining how to block image data, FIG. 3 is a diagram for explaining calculation of an evaluation matrix, FIGS. 4 and 5 are diagrams showing an example of an image of the evaluation matrix, and FIG. FIG. 7 is a diagram illustrating the calculation of parameters, FIG. 8 is a diagram illustrating the prior art, and FIG. 9 is a diagram illustrating each image quality defect. 10...ITV camera, 11...Light box,
12... Decoder, 13... Image processing device, 14.
...Display device. Applicant's Representative Patent Attorney Takehiko Suzue Figure 1 Figure 0 Figure 3 Molecule 1 Sheet Figure 2 (a) (b) (a) (b) Figure 0 Figure Figure 1 Figure (a) (b) (c) ( d) No. 91

Claims (2)

[Claims] (1) Divide the image data to be evaluated into multiple blocks in the matrix direction, find the average gray level for each block, and calculate the gray level by normalizing these block average gray levels by the average gray level in the row or column direction. A uniformity evaluation method characterized in that the evaluation target is evaluated by determining a block average matrix, and then determining each correlation between rows or columns of the block average matrix. (2) Divide the image data to be evaluated into multiple blocks in the matrix direction, find the average gray level for each block, and calculate the gray level by normalizing these block average gray levels by the average gray level in the row or column direction. A block average matrix is obtained, and then a matrix representing each correlation in each row or column direction of this block average matrix is obtained, and a uniformity parameter of each component in the diagonal direction of this matrix is obtained to evaluate the evaluation target. A uniformity evaluation method characterized by performing the following.