JP2007256245A - Appearance inspecting apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an appearance inspecting apparatus having the function of detecting an uneven defect and the function of detecting a pattern defect (minute defect) in a color filter or the like, from image data acquired by photographing a substrate such as the color filter or the like having a pattern shape by using an optics and a photographing system with high resolutions. <P>SOLUTION: The appearance inspecting apparatus comprises a photographing means 10 for photographing the original image of an object to be inspected such as the color filter or the like having a periodic pattern, a light projecting means 20 for projecting uniform light to the object to be inspected, a minute defect inspecting section 30 for determining a pattern abnormality, and an uneven defect inspecting section 40 for detecting an uneven stripe being horizontal or vertical to the object to be inspected. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an inspection apparatus having a function of detecting a periodic pattern shape abnormality on a glass substrate such as a color filter and a function of detecting transmission unevenness.

In the conventional unevenness inspection machine, the image is taken in a macro view from about a fraction of the periodic pitch. Therefore, the resolving power for inspection of fine defects needs to be about several tenths of the periodic pitch. When it is desired to detect unevenness and fine defects, it usually has to be added to the two optical systems, which requires cost and maintenance, and requires space.
In order to solve this problem, manufacturers try to handle micro and macro defects with a single optical system, but they do not function as detection of unevenness with pattern shapes and slight differences. Is the current situation.

In order to solve such a problem, a single inspection apparatus can detect a plurality of different types of defects from the same image data at high speed, and also reports the types of detected defects and comprehensively determines the defects. There has been proposed a defect inspection apparatus capable of performing (see, for example, Patent Document 1).
In the case of this defect inspection apparatus, it seems to be effective when the object to be inspected is plain with no pattern, but when it has a pattern shape like a color filter, the cause of the decrease in contrast is ultimately due to capturing the variation of the black matrix region It has become.
Japanese Patent Laid-Open No. 08-145907

Until now, in the inspection of a substrate having a pattern shape such as a color filter, the pattern defect (fine defect) inspection has been performed by acquiring an image having a high resolution, for example, a few tenths of a periodic pitch.
On the other hand, the unevenness inspection is performed using an image with a low resolution. That is, necessary original images are required for each, and if an inspection machine having the above two inspection functions is to be manufactured, it is necessary to prepare an optical system and an imaging system corresponding to the respective inspection functions.

  The present invention has been devised in view of the above problems, and pattern defects such as color filters (fine patterns) are obtained from image data obtained by imaging a substrate having a pattern shape such as a color filter using a high-resolution optical system and imaging system. It is an object of the present invention to provide an appearance inspection apparatus having a function of detecting a defect) and a function of detecting a mura defect.

  In order to achieve the above object in the present invention, first, in claim 1, an imaging means 10 for reading an original image of a test object having a periodic pattern such as a color filter, and uniform light is projected onto the test object. The light projecting means 20 for illuminating, a fine defect inspection unit 30 for determining a pattern abnormality, and a nonuniformity inspection unit 40 for detecting horizontal or vertical unevenness with respect to the object to be inspected are provided. This is an appearance inspection apparatus.

  Further, in claim 2, the two-dimensional image data acquired by the imaging means 10 is taken in by the fine defect inspection unit 30 and the unevenness inspection unit 40, and independently processed in parallel. This is an appearance inspection apparatus.

  In the present invention, the unevenness inspection unit 40 first calculates a luminance value and α indicating the maximum frequencies of R (red pixel), G (green pixel), and B (blue pixel) from the histogram of all data. The step of counting or counting normal values in the horizontal and vertical directions in the coordinates in the data excluding the brightness value ± α other than the luminance value ± α indicating the maximum frequency of each of R, G, and B, and movement for a period And a step of calculating an average and multiplying by a predetermined moving average, thereby having a function of determining a streak when a difference value from a predetermined moving average value exceeds a predetermined threshold. The appearance inspection apparatus according to 1 or 2 is provided.

  According to a fourth aspect of the present invention, the visual inspection apparatus according to any one of the first to third aspects is provided with a sensitivity correction optical filter 11.

  Furthermore, in claim 5, the periodic pitch acquisition means and the function of selecting a normal position candidate of the same position in the periodic pattern for the pixel of interest are provided. The appearance inspection apparatus described in 1. is used.

Here, the period pitch acquisition means is to acquire the vertical and horizontal period values of the black matrix from the design pattern file or the basic information list.
In addition, the pixel of interest is determined for each pixel whether it is defective or normal in order to examine where the defect is at the time of inspection. When the image data is expressed in the XY coordinate system, if the target pixel is I (i, j), the range of 0 ≦ i ≦ u and 0 ≦ j ≦ v is sequentially scanned. u and v are the numbers of pixels in the horizontal and vertical directions, respectively.
In addition, the function of selecting the same position normal part candidate in the periodic pattern of the target pixel is that the arbitrary number of points corresponding to integer multiples of the vertical and horizontal periodic pitches is the same position normal part for the target pixel. Get as a candidate value. This is a function that uses the median of the luminance values within the points listed as candidates as a reference value.

  By using the appearance inspection apparatus of the present invention, pattern defects (fine defects) such as color filters are obtained from the same image data obtained by imaging a substrate having a pattern shape such as a color filter using a high-resolution optical system and an imaging system. It becomes possible to inspect uneven defects with high sensitivity.

Embodiments of the present invention will be described below.
FIG. 1 is a schematic configuration schematic diagram showing an embodiment of an appearance inspection apparatus according to the present invention.
As shown in FIG. 1, the appearance inspection apparatus 100 includes an imaging unit 10 that images a test object having a pattern shape such as a color filter, a light projecting unit 20 that projects uniform light onto the test object, and an imaging unit. 10 includes a fine defect inspection unit 30 for determining pattern abnormalities that are captured and processed independently, and a nonuniformity inspection unit 40 for detecting horizontal or vertical unevenness with respect to the object to be inspected. The inspected object having a periodic pattern such as a color filter is transmitted and irradiated with uniform light without unevenness by the light transmitting means 10, and the transmitted light is imaged by the imaging means 10. The image data picked up by the image pickup means 10 is an appearance inspection apparatus that is processed by the fine defect inspection unit 30 and the unevenness inspection unit 40 to determine pattern abnormalities such as pattern defects (fine defects) and uneven stripes.

As the imaging means 10, an area camera, a line scan camera, or the like can be used.
The fine defect inspection unit 30 and the unevenness inspection unit 40 are each programmed and set in a computer, and perform control processing as needed to display the inspection result on a display, print out, and the like.

  According to the second aspect of the present invention, the image data picked up by the image pickup means 10 is distributed to the fine defect inspection unit 30 and the stripe unevenness inspection unit 40, and each of them is independently processed in parallel to determine a pattern abnormality. As a result, it is possible to reduce and maintain the inspection tact.

In the invention according to claim 3, in the unevenness inspection unit 40, a histogram is taken for the image data captured by the imaging unit 10, and normal ranges of R, G, and B are acquired and set, or Register as a known one in advance.
Then, binarization is performed at each position in the normal range where the luminance value ± α indicating the maximum frequency of each of R, G, and B.
That is, in the case of horizontal u pixels and vertical v pixels, normal uv (BYTE) is converted into 3 (u + v) (BYTE) data.

When the one-dimensional data of R, G, and B are respectively projected in the horizontal direction, 3v data arrays are acquired, and the number of stored data is counted as normal values.
Here, the binary data of R, G, and B are added in the horizontal direction to obtain an average value. That is, a process of project array data Prjt_u [j] = ΣI (i, j) / u, 0 ≦ i ≦ v is performed. However, 0 ≦ j ≦ v.
For R, G, and B, 3v data can be obtained by performing projection in the horizontal direction, and 3u data can be obtained by performing projection in the vertical direction in the same manner.

If there is a variation in the pattern of the periodic pattern, it cannot be processed as it is, so the processing is performed by adding the number of pixels overlapping the period. That is, when the 3v data array is divided into R array data, G array data, and B array data, an addition value corresponding to the pixel pitch is substituted for each.
Further, each projection data is subjected to a moving average for the pixel pitch.

  That is, when the pixel pitch is p, the data array Rv [i] = (Prjt_v [i] + Prjt_v [i + 1] + Prjt_v [i + 2] +... Prjt_v [i + P]) / P. However, the p pixels from Rv [vp] to Rv [v] are non-inspection regions. However, 0 ≦ i ≦ v. This process is similarly performed for Ru, Gv, Gu, Bv, and Bu.

Further, in most cases, the period does not become an integer multiple of the number of pixels pitch, so that noise is removed by further applying a moving average to correct this deviation. This is AveI ′ [i]. A difference between a nonuniformity cycle and a different moving average value, or a predetermined non-defective product standard frequency reference value that exceeds a predetermined threshold value is determined to be uneven. This is determined by performing a total of six times in the horizontal and vertical directions of R, G, and B.
Here, the moving average means that if the moving average number of moving pixels is m and the data array after moving average is AveI [i], 0 ≦ i ≦ u, AveI [i] = ΣI [k], i ≦ k ≦ i + m.
Further, if it is assumed that the width of unevenness occurs in n ± β pixels, a moving average of 2n or more is taken, and when the data array is AveI ″ [i], the difference is AveI ′ [i] −AveI ″ [i]. Yes, if the absolute value of the difference value exceeds the threshold value, that is, if Abs (AveI ′ [i] −AveI ″ [i])> Threshold, it is determined that the stripe is uneven.

In an object to be inspected such as a color filter, Y (yellow), M (magenta), and C (cyan) may be added in addition to the R, G, and B filter patterns. In this case, the above procedure can be expanded and processing can be performed although the processing load is twice as much as possible.
In this way, the object to be inspected, such as a color filter having any color filter, is the target filter color.

  In the invention according to claim 4, the histogram may not be clearly separated for R, G, B, Y, M, and C depending on the spectral sensitivity characteristics of the light source and the camera. In addition, although separation is possible, when the band regions of R, G, B, Y, M, and C are distributed over a wide range, the detection ability of each color is different in the pattern inspection. By designing and mounting an optical filter that corrects this, image data that can be used for both unevenness inspection and pattern inspection can be obtained by correcting to an appropriate distribution.

According to a fifth aspect of the present invention, the periodic pitch can be read from design data, recipe, or image data.
Gets the vertical and horizontal period values of the black matrix from the design pattern file or basic information list. The periodic pitch is set at the time of design.
When the periodic pitch is t and the camera resolution is r, the horizontal pixel pitch Ph = t / r and the vertical pixel pitch Pv = V / f are obtained.
However, V is a subject moving speed, and f is a sampling frequency of the camera.
Therefore, the same position of the periodic pattern with respect to the inspection target pixel is the position of nPh in the horizontal direction and nPv in the vertical direction. An arbitrary point is selected from this position, and the difference from the target pixel is calculated using the median value or the average value as a reference value. take. n is an integer.

Hereinafter, a method for inspecting stripe unevenness and pattern defects (fine defects) using the appearance inspection apparatus of the present invention will be described.
In the appearance inspection apparatus 100 of the present invention shown in FIG. 1, first, uniform illumination is performed by the light projecting means 20 from the lower surface of the inspected object 71 having a periodic pattern such as a color filter.
Next, the color filter of the inspected object 71 is imaged by the imaging means 10 composed of a line camera to obtain the color filter image shown in FIG.
Here, the optical characteristic correction filter 11 is attached to the front surface of the lens of the imaging means 10 to correct the sensitivity, and the luminance values of R, G, and B are adjusted to a level that can be separated as shown in the histogram of FIG.

Next, FIG. 3 shows an example of a histogram of the color filter image picked up by the image pickup means 10. The horizontal axis is the luminance value, and the vertical axis is the frequency.
The histogram in FIG. 3 represents the luminance distribution of the color filter image. The horizontal axis represents the luminance value, and the vertical axis represents the frequency. The peaks of R, G, and B are separated as shown by 51, 52, and 53, respectively. The 54 is a black defect such as a noise component or foreign matter, 55 is a BM (black matrix), and 56 is a noise component or white defect.
The acquired two-dimensional image data is taken into the fine defect inspection unit 30 and the stripe unevenness inspection unit 40 via the data buffer, and is independently processed in parallel.

In the non-uniform stripe inspection section 40, focusing on R in FIG. 3, only the normal value region of the R peak ± α is binarized into one-dimensional data is the image in FIG.
When this is projected in the u and v directions, the frequency of the abnormal line drops, and the point is depressed. The moving average is taken for the pixel pitch, and the pixel pitch and the pattern period error are added as noise components, so the waveforms obtained by multiplying and removing the moving average become Pjct_u and Pjct_v, respectively, and can be expressed as the stripe unevenness 61.
Here, taking a projection in the u and v directions means adding binary data of R, G, and B in the horizontal u direction to obtain an average value. That is, the process of project array data Prjt_u [j] = ΣI (i, j) / u, 0 ≦ i ≦ v is performed. However, 0 ≦ j ≦ v.
Similarly, the vertical direction of the v direction is Prjt_v [i] = ΣI (i, j) / v, 0 ≦ j ≦ u.
Taking the moving average for the pixel pitch means that the number of moving average moving pixels is m, and the data array after moving average is AveI [i], where 0 ≦ i ≦ u. , AveI [i] = ΣI [k], i ≦ k ≦ i + m.

  Accordingly, if it is assumed that the unevenness width occurs in n ± β pixels so as not to overlap the unevenness period, a moving average of 2n or more is taken for this waveform, and this difference value (AveI ′ [i] and AveI ″ [i ]) Exceeds a threshold value, it is determined as a defect. Alternatively, a specified value is determined, and when the difference from the constant reference value exceeds the threshold value, it is determined as a defect as an abnormal value.

The fine defect inspection unit 30 acquires, as a reference candidate value, an arbitrary number of arbitrary points corresponding to n times the vertical and horizontal periodic pitches as the same position normal part for the target pixel. The luminance values in the points listed as candidates are rearranged, and the median value is used as a reference value.
When the difference between the target pixel and the reference value exceeds the threshold value, the target pixel is regarded as a defect.
The method of taking a plurality of points is improved in accuracy because the conveyance direction and the vertical direction are not affected by the subject speed fluctuation.

It is a schematic block diagram which shows one Example of the external appearance inspection apparatus of this invention. 4 is an explanatory diagram illustrating an example of a color filter image captured by an imaging unit. FIG. 4 is an explanatory diagram illustrating an example of a histogram of a color filter image captured by an imaging unit. FIG. FIG. 4 is an explanatory diagram in which only the normal value region is binarized and displayed as one-dimensional data, paying attention to R in the histogram of FIG. 3.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Imaging means 11 ... Optical filter 20 for sensitivity correction ... Projection means 30 ... Fine defect inspection part 50 ... Unevenness inspection parts 51, 52, 53 ... Selection area 54 ... Noise components, foreign substances, etc. Black defect 55 ... BM (Black Matrix)
56 …… Noise component and white defect 61 …… Straight irregularity 71 …… Inspection object 100 …… Appearance inspection device

Claims (5)

  An image pickup means (10) for reading an original image for inspecting a minute defect of an inspection object having a periodic pattern such as a color filter, a light projection means (20) for projecting uniform light to the inspection object, and a pattern abnormality An appearance inspection apparatus comprising: a fine defect inspection unit (30) for determining; and a non-uniformity inspection unit (40) for detecting horizontal or vertical unevenness with respect to an object to be inspected.   The appearance inspection according to claim 1, wherein the two-dimensional image data acquired by the imaging means (10) is taken in by the fine defect inspection unit (30) and the unevenness inspection unit (40) and independently processed in parallel. apparatus.   The unevenness inspection unit (40) first calculates or sets the luminance value and α indicating the maximum frequencies of R, G, and B from the histogram of all data, and the luminance value indicating the maximum frequencies of R, G, and B, respectively. In the coordinates in the data excluding ± α, the number of normal values is counted and counted in the horizontal and vertical directions respectively, and the step of calculating the moving average for the period and multiplying it by the predetermined moving average The visual inspection apparatus according to claim 1, further comprising a function of determining a streak when a difference value from the moving average value exceeds a predetermined threshold value.   The appearance inspection apparatus according to any one of claims 1 to 3, further comprising an optical filter (11) for sensitivity correction.   5. The appearance inspection apparatus according to claim 1, further comprising: a periodic pitch acquisition unit and a function of selecting a normal position candidate of the same position in a periodic pattern of a target pixel.