TWI354100B - Unevenness inspecting method, unevenness inspectin - Google Patents

Description

1354100 IX. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention relates to a technique for inspecting streak Unevenness on a substrate. [Prior Art] V In the prior art, when a pattern of a panel is formed on the main surface of a glass substrate for a display device, an anti-money agent liquid is applied on the main surface to form an anti-silver agent film. When a resist film is formed, for example, if a minute unnecessary substance exists on the substrate before the application of the resist liquid, the unnecessary object may hinder the gap of the resist liquid used for the soil, in the resist The film on the film extends from the position of the unnecessary object to cause streak unevenness. In recent years, it has been possible to improve the manufacturing process, the formed pattern, and the like by detecting the existence position of such streak unevenness on the substrate (or by detecting unevenness of the stripes). As an example of the method of detecting the position of the unevenness of the stripe, the method disclosed in the Japanese Patent Laid-Open Publication No. 2005-345290 (the document discloses that the individual pixels in the image to be inspected are sequentially focused on the pixel. In an area of a predetermined size centered on a 5 megapixel, a plurality of pixels whose values are above the threshold and an angle reference line passing the pixel are In the meantime, the sum of the distances is calculated, and based on the minimum value of the sums, the value is used as the value of the pixel of interest, and an image indicating the existence position of the unevenness of the streaks is obtained. The second disclosed in Japanese Laid-Open Publication No. 6-105884 is to photograph the substrate, apply a low-pass filter treatment to the obtained image, and use the 312XP/invention specification for high-pass filtration treatment when filtering through the filter. Replenishment) /96-08/96116068 ^ 1354100 Figure ^ group 'restricted to the pixels in the area of the high-speed window, so as to improve the inspection of the edges of each area on the substrate with multiple corners In addition, the method of recent years is based on the degree of unevenness of the substrate quality (hereinafter referred to as "intensity of unevenness"), and it is judged whether or not it is allowed (4), V is in Japanese Patent Laid-Open No. 9- The technique disclosed in Japanese Patent Publication No. 68502 is to multiply the values of the pixels perpendicular to the direction at the respective positions in the predetermined direction for the image indicating the shadow mask, and to obtain the multiplication data. The value of the material is divided by the value corresponding to the smoothed data obtained by smoothing the multiplication data, and based on the obtained normalized data, it is determined whether or not the unevenness of the shadow of the shadow mask is allowed (that is, whether it is a defect). The method proposed in Japanese Laid-Open Patent Publication No. Hei 10-197451 is to divide the value of each pixel of the image obtained by the object by the value of the pixel corresponding to the image obtained by smoothing the image. Obtaining a normalized image for the difference between the average value of the pixel values in the normalized image and the pixel whose absolute value is below the predetermined threshold # value, used to generate a new image, the pixel of the image value Row multiplication is used to determine whether or not the unevenness is allowed. However, when a plurality of panel patterns are formed on the substrate (that is, the substrate is multi-dehorned), the plurality of panels respectively correspond to the display regions of the panel. The rectangular area 'is set to be vertically and horizontally arranged with a gap. In this case, the same pattern is formed in a plurality of rectangular areas, but the gap between the five-shaped areas is formed by various patterns, so When the image obtained by the substrate is used in the original state in the uneven inspection, the area corresponding to the rectangular area often causes a false signal. Therefore, when the unevenness is checked, the 312XP>^ instructions (supplement)/96·08 /96116068 η ' 1354100 Preferably, in the image representing the substrate, the area corresponding to the rectangular area is added, and the pixel value is added (that is, the area is used as a mask) = the image of the object to be inspected. However, when the stripe is not uniform for such an image, when there is a uniformity in which a plurality of rectangular regions are covered in the substrate, the same stripe is uneven, and the source is uneven in the V-shaped region. Most of the stripe shapes detected by the Lieutenant General are considered to be independent of each other, so it is impossible to effectively specify the cause of uneven streaks. In addition, in recent years, in order to analyze the unevenness of the stripes on the substrate, or to determine whether the stripe of each of the areas of the cutting unit of the slab is good, etc., it is required to accurately check the existence position of the stripes on the county board, and individually Obtaining the degree of influence of the unevenness of the ^ and the text on the quality of the substrate (hereinafter referred to as "the strongness of the unevenness of the stripes, but in the above method 1", it is necessary to use multiple angles when the unevenness of the fineness is required with good precision. A very large amount of calculation is performed on the baseline. In addition, when the evaluation value of the unevenness of the stripes is required, it is necessary to recalculate: calculus. Therefore, it is necessary to use a novel method to detect the unevenness of the streaks with good precision and to easily obtain the intensity of unevenness of the stripes. In addition, in addition to the unevenness of the above-mentioned stripes on the substrate, there are also defects such as unnecessary objects which are the cause of unevenness of the stripes. In addition, there are other defects that do not have important characteristics. In the case where the defects overlap (that is, in the case where the regions indicating the defects in the target image of the substrate overlap), it is difficult to distinguish the defects. 】 The present invention is directed to a film thickness unevenness inspection method for inspecting unevenness of stripes on a multi-angled gusset in a plurality of rectangular regions, and the object thereof is to easily 312 ΧΡ / invention specification (supplement) / 96- 08/96116068 8 13541 〇〇 ;; / / letter cover a plurality of rectangular areas and extended stripes are not about film thickness unevenness inspection method includes: = ° (a) For the substrate photography, the obtained gaps F ^ The original image of ^ 5 corresponds to the area of the 丨糸 遮 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , The ratio of the width of the width direction to the length direction is a predetermined value step area as a plurality of stripe unevenness element regions (.) From among the plurality of stripe unevenness element regions, the length direction is detected as being arranged in the length Direction, and two stripe uneven element regions existing in adjacent rectangular regions. According to the present invention, stripe unevenness covering a plurality of rectangles can be easily detected. In a preferred embodiment of the present invention, b) in the step, by representing the unevenness of the area In the (c) step, the stripe unevenness element line segment is used as the corresponding stripe unevenness element area in the (c) step, and it is possible to more easily detect and cover a plurality of rectangular areas. In another preferred aspect of the present invention, the step of displaying the position of the connection region connecting the two stripe unevenness elements to the display portion is further provided. It is confirmed that the stripe unevenness is extended by covering a plurality of rectangular regions. In one aspect of the present invention, the step (b) includes: (bi) binarizing the target image by using a plurality of thresholds for obtaining With a plurality of thresholds 312XP / invention manual (supplement) / 96-08/96116068 9 1354100 ✓ η should be a plurality of binary image steps (10)) in multiple = one, the length in the length direction And a step perpendicular to the longitudinal direction and a step of forming a stripe region having a value equal to or greater than a threshold value, and (10), in which two stripe regions adjacent to each other are adjacent to each other, and the stripe regions overlapping each other are grouped and included In the same regional group, get In the case where the stripe is not at least one region group, the step of specifying a plurality of stripe unevenness element regions is further included in the method for inspecting the film thickness unevenness (4) for at least i region groups, and each of the regions is obtained. There is a step of the L-limit value |a of the binary image of the stripe region, or the sum of the length or the area of the stripe region as the unevenness of the stripe. In this way, the strips can be detected with good precision, and the text can be unevenly distributed, and the intensity of streak unevenness can be easily obtained. In another aspect of the present invention, the step of detecting the streak unevenness region indicating the unevenness of the stripe is detected from the object image by performing the steps (a) to (c), wherein the stripe uneven region is in the length direction Length and perpendicular to the length: the ratio to the width is above the established value; the film thickness unevenness inspection method is further included. (g) The object image is binarized to detect the defect to be screened. a step of the defect to be screened indicating that there are other defects in the image of the object to be screened; and (h) when the defect to be screened is overlapped with at least a portion of the uneven area of the stripe, the defect is to be screened The first evaluation value of the defect strength of the region, and the second evaluation value regarding the unevenness of the stripe unevenness in the stripe region are used to determine whether or not the defect-to-screening region is to be removed. In this way, when the defect-to-screening area overlaps with at least a portion of the stripe unevenness area, it is possible to easily determine whether or not the defect-to-screening area is to be removed, and as a result, the unevenness of the stripe and the defect can be correctly detected. To be screened 312XP / invention manual (supplement) / 96·08/96116068 10 1354100 area. The present invention is also applicable to: a film thickness unevenness inspection device that checks unevenness of stripes on a substrate of a plurality of rectangular objects in a rectangular object; and a computer-readable recording medium on which a program can be recorded on a computer Across multiple rectangular areas, the unevenness of the stripes on the multi-dehorned substrate is checked. The above and other objects, features, aspects and advantages of the present invention will become apparent from [Embodiment] Lutu 1 shows a film thickness unevenness inspection apparatus according to a first embodiment of the present invention! Construction. Uneven thickness inspection device! In the glass substrate 9 used for a display device such as a liquid crystal display device, a coating device such as a spin coating method or a slit coating method is used, and a resist liquid is applied onto one of the main faces 91 to obtain a formed pattern. The image of the anti-surname film 92 checks the unevenness of the stripes on the film 92 of the substrate 9 based on the image. Further, on the substrate of the present embodiment, a pattern of a plurality of face plates for a display device is formed (or is formed in advance). Here, the unevenness on the substrate 9 means a region of a certain area or more which is specified in accordance with the local light and dark fluctuation (but usually the boundary of the region is not clear), and the unevenness of the stripe is defined as the length α of the region in the longitudinal direction, and The ratio (α/々) of the width yS perpendicular to the longitudinal direction is at a predetermined value two. Of course, if the condition can be satisfied substantially, the definition of the streak unevenness can be appropriately changed, and other conditions can be added. As shown in Fig. 1, the film thickness unevenness detecting vibration i is provided with a carrier α 2, and the main surface 91 on which the film 92 is formed (hereinafter referred to as "upper surface 91") is oriented toward the upper side 312 ΧΡ / invention manual (repair) /96-08/96116068 11 1354100% / (the (10) side in Fig. 1) is used to hold the substrate 9; the light irradiation portion 3 is incident on the substrate to be held on the substrate 9 held on the stage 2 The film is used to receive the light reflected from the film 92 on the upper surface 91 of the substrate .9 by the light irradiation portion 3; the moving mechanism 21 is used to illuminate the stage 2 with respect to the light. The portion 3 and the light receiving unit 4 are moved; and the computer serves as a control unit of the film thickness unevenness inspection device 1. The surface of the (+Z) side of the stage 2 is preferably treated with a black gloss finish. The moving mechanism 21 has a structure in which a bolt (not shown) is connected to the motor 211, and the rotation of the motor 211 causes the stage 2 to move along the upper surface 91 of the substrate 9 in the X direction in FIG. The light-irradiating portion 3 is provided with a light source for emitting white light (that is, light having a skin length including all of the visible regions), and a cylindrical quartz rod 32 perpendicular to the stage 2 In the direction i of the moving direction, the extension of the 'Cylinder lens 33' extends in the Y direction. In the light-irradiating portion 3, the halogen lamp 31 is applied to the end portion of the quartz rod 32 at the (+ γ) side, and the light incident from the halogen lamp 31 to the quartz rod 32 is converted into linear light extending in the γ direction (also In other words, the beam cross section is linear in the γ direction, is emitted from the side surface of the quartz rod 32, and is guided to the upper surface 91 of the substrate 9 via the cylindrical lens 33. In other words, the quartz rod 32 and the cylindrical lens 33 are optical, and the light from the halogen lamp 31 is converted into linear light 'which is perpendicular to the moving direction of the stage 2' and guided to the upper surface g1 of the substrate 9. In Fig. 1, the optical path from the light-irradiating portion 3 to the substrate 9 is indicated by a one-dot chain line (the optical path from the substrate 9 to the light-receiving unit 4 is also the same). A portion of the light emitted from the light-irradiating portion 3 is reflected by the 臈92 on the upper surface 91 of the substrate 9 on the (+Z) side of the 312XP/invention specification (supplement)/96-08/96116〇68 12 1354100. The film 92 is light-transmitting to the light from the light-irradiating portion 3, and the light from the light-irradiating portion 3 is not reflected by the upper surface of the film 92, and the permeable film 92 is reflected by the upper surface of the substrate 9 (that is, the lower surface of the film 92). . The interference light of the film thickness unevenness inspection device 1' reflected on the upper surface of the film 92 of the substrate 9 and the light reflected on the upper surface 91 of the substrate 9 is incident on the light receiving unit 4 via the filter 43 and the lens 42. The interference light that is a predetermined wavelength is guided to the imaging unit 41. The imaging unit 41 is provided with a line sensor having a plurality of light receiving elements (for example, CCD (Charge Coupled Device)) arranged in a line in the γ direction, and the interference light from the substrate 9 is received by the line sensor to obtain the intensity of the interference light. Distribution (that is, the output value from each light-receiving element in the γ direction = cloth). Actually, as the substrate 9 moves toward the x direction, the line sensor of the photographing portion 41 repeatedly obtains the intensity distribution of the interference light for obtaining the secondary image of the film 92 on the substrate 9. As shown in Figure 2, the computer 5 is a general computer system. The connection is: ^PU51 is used to enter various calculation processing; _52 is used to memorize the basic process, and i RAM53 is used to read various information. The bus line is appropriately connected via an interface (I/F), etc.: Figure _ 疋 疋 CD 54 for memorizing information; Page 5 5 5 'Becoming a display unit - 々 一 贝 贝 峥There is no variety of poor news; keyboard 56a and mouse 56b (hereinafter collectively referred to as "钤λ _ r, 轸 入 · Φ ^ ”), used to accept from the operation::: enter, take ... 7, for reading The information from the CD, the disk, the optical disk, etc. can be read from the _ _ 17 'pro-media 8 information; and the communication unit 58, connected to the film thickness unevenness inspection device 丨 clothing straight 1 other components. Before the computer 5', the program 312XP/invention manual (supplement)/96-08/96116068 13 丄354100 54 is read from the recording medium 8 via the Jing% # reading device 57 and stored in the fixed optical disc 54. Then, the program 5::RA=' and the CPU 51 execute the program according to the program in _3, which is executed by the computer, and is used to check the stripe on the substrate 9 by using the ^. The operation of the calculation unit is shown in FIG. The cpu 51 is used to display the cpu 51 in accordance with the program (4), the r7 CPU 51, 52, the RAM 53, the fixed optical disk 54, and the like. In FIG. 3, the target image acquisition unit 62 in the calculation unit 6 is realized by the CPU 51 or the like, and the stripe unevenness element/, The functions of the unevenness detecting unit 64 and the display control unit 65. In addition, these functions can also utilize dedicated circuits exclusively. It can also be realized by the circuit. It can also explain the use of the unevenness of the film thickness (4) for the scratch detection of the unevenness of the stripe. Fig. 4 is a view showing a flow of processing in which the film thickness unevenness inspection device i checks the unevenness of the stripes on the base. When the stripe on the inspection substrate 9 is not: 'First' the substrate 9 is held on the stage 2 located at the start position of the inspection 幵1 shown by the solid line in Fig. i' and then the stage 2 is oriented (+ magic direction) Then, the linear light that is emitted from the light irradiation unit 3 and incident on the upper surface 91 of the substrate 9 at a predetermined incident angle is irradiated onto the linear irradiation region on the upper surface 91 (hereinafter referred to as "linear irradiation region". ”, the linear irradiation region is relatively moved to the substrate 9. The light from the light irradiation unit 3 is reflected on the upper surface 91 of the substrate 9, and the interference light is guided to the imaging unit 4 to receive light by the line sensor, and the substrate is taken. The intensity distribution of the interference light in the upper-line irradiation region is 9. The output of each of the light-emitting elements from the line sensing is converted into a value of, for example, 8 bits (of course, other than 8 bits) according to a predetermined conversion formula (pixel value). , 312XP / invention manual (supplement) / 96-08/96116068 14 1354100 and sent to the computer 5 β, both check the device 1, when the stage 2 moves in the (+ Χ) direction of the thousand wells The movement is synchronously 'repeating the intensity of #你1" of the photographing unit 41 The pixel value is output to the computer 5. Then, when the #2 is moved to the inspection end position, the movement mechanism 21 = the heart 2_ is stopped, and the imaging unit 41 captures the entire base surface 92 and acquires more. The 2nd-dimensional image of the color tone (the image before the application is called the "original image") is input to the calculation unit 6 of the electronic device 5 (step S11). Then, the image of the object in the calculation unit 6 The production is compressed into the first] and Germany. Production LL + Earth pressure king 1 set you ~ Bao ^ Here the original image is located at the coordinates (X, Y) = prime pixel value is represented by Fxy, to Sa map The range of the prime xsa circle is = the original image of the original image, and the pixel value Axy of the main eye element at the coordinate (M) is obtained by the formula (丨).

Sa(x+1) s.(y+l) -1 Σ _Σ fxy/i Αχ% :S«x :sey (1) In the present embodiment, Sa^ 4 (pixel) is used, so the equation (1) is used. The difference can increase the S/N ratio of the first image to four times that of the original image. After the 1 image (pressing the U image), the first image is subjected to low-wave processing to suppress the influence of the high-frequency noise from the first image, and the second image is generated. Window for the calculation range of the low-pass filter processing of the shirt = 1 square length (2S1 + 1) square of the pixel 'the pixel value Lxy of the attention pixel of the (x, y) of the second image, using the attention map The pixel value A (see equation (1)) of the first image of each element in the vicinity of the element is obtained by the formula (2). 312xp/invention manual (supplement)/96-08/96116068 15 • · · (2) 1354100

Axy/(2si + 1)2 = The post-image is subjected to high-pass chopping processing, and is removed from the second image: a low-frequency density variation that hinders contrast enhancement processing described later 5 third image. Here, the pixel value of the second image of each pixel in the vicinity of the pixel (x, y) (see equation (2)) is obtained by the equation (3). • H-=W( Σ Σ Lx,/(2s2-fl)2) X=x—S2 Y=y—33⁄4 ' · · · ( 3 ) Equation (3) represents the calculation range used to determine the high-pass filtering process The window is such that Z is centered on the pixel of interest, and the length of each side is (the case of the square of the square of the pixel. As described above, in the object image generating unit 6i, the third image compressed for the original image is After the low-pass filter processing is applied, 'the high-pass chopping process for performing the predetermined spatial frequency band is performed by applying the high-pass filter processing (step S12). ^ The contrast-emphasis processing is performed on the third image in the target image generating unit 6 An emphasized image is generated (step S13). The pixel value Exy of the pixel of interest of the emphasized image coordinate (X'y) uses the pixel value H xy ' contrast coefficient rc of the attention pixel of the third image, and the background value b, obtained by the formula (4). In the present embodiment, !-. is 〇. 〇1, 〇〇2, 0.05 or 〇. 1, b is 127.

Exy = b (( Hxy — 1 ) / rc + 1) · (4) Here, a plurality of rectangular regions (hereinafter referred to as "rectangular display regions") are arranged on the upper surface 91 of the substrate 9 with the gaps arranged vertically and horizontally Corresponding to the display area on the panel of the display device and having the same pixel, the substrate 9 becomes 312XP/invention specification (supplement)/96-08/96 U 6068 16 rectangular areas are more than chamfered (ie, in A plurality of parts (panels) The second image generating unit 61' adds a background value b to the area where the emphasized image corresponds to the base and the outer space (hereinafter referred to as "background area", as shown in Fig. 5 - m + female ^ ^ ?1Q, . ^ No, used to generate and prepare the background G field 719 (in Figure 5, the new image is attached with additional parallelism (as described later in the 'Because of the wind and shoulders. 卩6 fine plus The following processing object: image, hereinafter referred to as "target image" 71 (step (10). In addition, the medium multi-tone object image 71 is displayed in a simplified manner, in fact, = medium = display area i 7U and background area 719 The realm is not - the average concentration of the target image 71 (that is, the average value of the pixel values) becomes about 12 7 . When the target image 71 is generated by the living unit 61, the target image is binarized by the binary value=the respective values 103 to 1223 which are smaller than the background value m, and the target image is binarized. = the value of the pixel and each threshold (hereinafter referred to as the "lower threshold"). In the second year, the value of the second is below the threshold, with "1" added. The value of the pixel, with the addition of "0", is used to obtain 20 binary images corresponding to the lower limit value. For the Ί 1 7 0# Μ table is not a binary image 'Figure 6A shows the next limit When the value is 2, 隹2 2nd image, FIG. 6B shows that the lower threshold is set to 112. ^Fig. 6C shows that when the lower threshold is 1G3, 2 is lower than the value of the moonscape value. The threshold value is obtained by 2 = = in the process, as shown in Fig. 6A to Fig. 6c, with the lower threshold value k being small (the background value is 127), the median value of the binary image is obtained. (Supplement) /96〇8/96n_i7 Ϊ354100 prime (the number of white pixels in Fig. 6A to Fig. 6c 47) is small. Then, in the binary image acquisition unit 62, each value is larger than the background 132. ~151 as a threshold The value (hereinafter referred to as "the upper threshold value is greater than the upper limit value of the target image, and the pixel is added to the upper limit value. The pixel is added to the threshold of the threshold to add the value of 132 to 151. Respectively: should be 2 0 2 ϋ 丨 丨 I gold. J to Linyi μ ~ like in the above, take the value greater than the background value as the upper:: value to obtain the binary image of the above process, with : a small value? 6 from the background value 127) The method of the pixels in the binary image is described in the binary image acquisition unit 62. The number of values is 2, and the target image is binarized. Obtain a binary image with a plurality of thresholds = two solids (40 in the present embodiment) (step > 'when the threshold value and the upper threshold value are used, Take: The value of the area corresponding to the background area in the binary image becomes 0. In the case of the parent, the two-striped region-specific portion 631 is a set of pixels that are continuous in each binary image (the τ is called "closed = μi k processing target, and the predetermined area is removed. (Picture number):: Area: 'Remove it from the binary image. Then, calculate the direction (angle) of each of the closed axes used to determine the inertia axis, and find the closed S domain along the inertia The circumscribed rectangle of the ancient a U ^ 。. In addition, it is usually the direction of the length of the closed region. In the following description, the 'inertia spindle is the length along the closed region. Figure 7 shows the binary image. Multiple closed areas n. 312XP / invention manual (supplement) / 96-08/96116068 18 1354100 = c. In Figure 7t 'each closed area 721 721. Connected to the rectangle with 'mouth has the number 723a ~ 722c The thin line indicates that in the stripe region specifying portion 63!, the respective closed regions 721a to 7(1) are connected to the rectangle 7n I, and the ratio of the length of the principal axis of the inertia to the width π of the direction perpendicular to the principal axis of the inertia (L1/ W1), compared with the established value (for example, 2, only this The closed region 72ia, which is a predetermined value or more, is a stripe region (hereinafter, the same symbol as the closed region (2) core (10) is added), and the other closed region 721c is removed from the binary image. In this manner, in the stripe region The specific portion 631 calculates a plurality of binary images so that the moments of the respective closed regions are used for the length of the specific length direction and the ratio of the two sides perpendicular to the longitudinal direction, and the ratio of the squares is equal to or greater than the value of the stripe region. (Step S16). In addition, the closed area of the binary image becomes larger when the lower limit value or the t-proximity value approaches the background value, so the binary image detection at a certain lower limit value or upper limit value is In the striped area, the binary image near the threshold value or the upper threshold value may have a closed area with a small aspect ratio, and is not limited to detecting the stripe area. In addition, in each of the stripe areas 721a, 721h, , s, a^·

The Ulb is supplemented by the center of gravity 723a, 72 in the main axial extension of the inertia, and the end points are set on the circumscribed rectangle 72h, the line segment on the side of the shackle (indicated by a dotted line in Fig. 7), which is specified to represent the stripe The stripe area line segments 724a, 72 of the regions 721a, 721b. Then, the stripe area updating unit 6-32' of the stripe unevenness element specificity 63 focuses on a plurality of areas corresponding to the plurality of rectangular display areas 7K on the substrate 9 with reference to 5) (hereinafter Also referred to as a rectangular display area in the rectangular display area (hereinafter referred to as "Note 312XP/Invention Manual (Replenishment)/96-〇8/96116068 19 1354100-shaped rectangular display area"), a specific attention rectangle display Any one of the plurality of stripe region line segments included in the region as a specific stripe region line segment. Then, between the line segments of the specific stripe region and the other stripe region segments in the display area of the eye-catching rectangle, it is determined whether or not the line segments of the stripe regions can be connected. • Fig. 8 is a diagram for explaining the determination of whether or not the stripe region line segment can be connected. In Fig. 8, it is emphasized that the slopes of the stripe region line segments 724a and 724b in Fig. 7 are different from each other. When the line segment area line segment 724a is used as the specific stripe area line segment 'determination and the connection between the stripe area line segment 724b, the center of gravity 723a of the stripe area 721a represented by the characteristic stripe area line segment 724a is first obtained (may also be the stripe area line segment) The midpoint, hereinafter, the connection line of the center of gravity 723b of the stripe region 721b represented by the stripe region line segment 724b (indicated by a broken line with the symbol R1 attached in Fig. 8). Then, the distances D1, D2 between the end points 725a, 726a of the specific stripe region line segment 724a and the straight line R1, and the distances j) 3, D4 between the end points 725b of the stripe region line segment 724b, the 726b and the straight line R1 are confirmed ( That is, all portions, such as the length from the end points 725a, 726a '725b, and 726b to the perpendicular of the straight line Ri, are all below the predetermined first threshold. When any one of the distances D1 to D4 is larger than the first threshold, the connection of the stripe region line segment does not hold. When the distances D1 to 4) are all below the first threshold, the extending direction of the specific stripe region line segment 724a (the longitudinal direction of the stripe region 721a indicated by the specific stripe region line segment 724a is hereinafter referred to as "longitudinal direction") The length direction of the stripe region line segment 724b is substantially the same (refer to FIG. 7). Then, the specific stripe region line segment 724a and 312XP/invention specification (supplement)/96·〇8/96116068 20 1354100 stripe region line segment 724b are obtained. The distance between the closest end points 726a, 725b and the sum of the lengths of the specific stripe area line segment 724a and the stripe area 724b. When the ratio of the sum of the distance D5 and the length is greater than the predetermined second threshold, the stripe area line segment The link is not established. When the ratio is below the predetermined second threshold, the 'special stripe region line segment 724a and the stripe region line segment 724b are close to each other' and then find the farthest between the specific stripe region line segment 724a and the stripe region demarcation line segment 724b. The distance D6 between the endpoints 725a, 726b, and the ratio of the length sum of the particular stripe region line segment 724a to the stripe region line segment 724b. Then, when the ratio is greater than or equal to the predetermined third threshold, the specific stripe region line segment 724a and the stripe region line segment 724b are arranged in the same length direction, and can be connected. When the third margin is not reached, the stripe region line segment is The link is not established. Then, the specific stripe region line segment 724a and the stripe region line segment 724b that are permitted to be linked are connected. Specifically, the specific stripe region line segment 72 is cut at its midpoint into two broken line segments, and the stripe region line segment 724b is split at its midpoint into two broken line segments, and the torque of the set is calculated. The direction of the inertial main pumping is obtained for the four broken line segments, and a straight line extending in the direction of the inertia main axis and passing through the center of gravity of the four broken line segments (that is, the 'inertia spindle') is obtained as shown in FIG. The dotted line with the symbol R2 is attached (however, the dotted line is partially thickened). Then, the intersection point of the perpendicular line falling from the end of each of the specific stripe region line segment 724a and the stripe region line segment (10), 726a, 725b, 726b to the straight line R2 and the straight line R2 is obtained, the intersection point which is the farthest from each other among the four intersection points, (4) The line segment as the end 312XP/invention specification (supplement)/96.08/96116068 1354100 point is obtained as a new stripe region line segment (the portion of the thick broken line in the straight line R2 indicated by a broken line in Fig. 9), and two stripes are obtained. The area line segments 724a, 724b are updated to be one stripe area line segment. Actually, in addition to the decision to permit the entire stripe area line segment to be joined to the specific stripe area line segment 724a, the connection of the specific stripe area line segment 724a and the stripe area line segments is simultaneously performed. That is, each of the plurality of stripe region line segments 724a and the plurality of stripe region line segments that are permitted to be joined is used to obtain a plurality of break line segments, and the set of the break line segments is obtained. The inertia main axis is lowered from the end point of the specific stripe region line segment 724a and the end point of the plurality of stripe region segments to which the connection is permitted to the perpendicular line of the inertia main axis, and the intersection points of the inertia main axis, and the two intersection points which are the most distant from each other are used as the end points. The line segment is a new stripe region line segment, and the specific stripe region line segment 724a and the plurality of stripe region line segments that are permitted to be connected are updated into one stripe region line segment. When determining whether or not the specific stripe region line segment 724a is connected to other stripe region_line segments, the stripe region line segment that is updated from the specific stripe region line segment 724a in the hatched rectangle display region is specified (special stripe when not connected) One of the stripe area line segments other than the area line segment 7 2 4 a) is used as a specific stripe area line segment to determine and connect with other stripe area line segments. Actually, in each of a plurality of binary images, each rectangular display area is used as a horizontally-shaped rectangular display area, and the above-described processing is performed to make the longitudinal direction of the plurality of stripe area segments specified in each rectangular display area the same. In the length direction (that is, arranged in the length direction to be aligned on the same line), and a plurality of stripe regions 312XP/invention manual (supplement)/96-08/96116068 22 1354100 line segments that are close to each other are updated into one stripe region In the line segment (step si), the threshold value of the region group acquisition unit in the stripe unevenness element specifying unit 63 is adjacent to each other (in the present embodiment, the threshold value is continuous). In the hexadecimal image, it is determined that the square of the stripe area and the stripe area of the other one of the rectangular display areas corresponding to the rectangular display area including the stripe area line segment Fig. 1G overlaps and indicates that the stripe region line segments 724d and 72 included in the two binary images adjacent to each other in the binarization limit are cut. When the domain line segments 724d and 724e can be grouped, first, a straight line R3 (wherein the portion extending from the stripe domain line segment 724d is extended by a broken line) for extending one stripe region line segment 724d is set, and the other stripe region line segment 724e is obtained. The distances D7 and D8 of the two end points 725e and 726e are respectively from the straight line R3. When the distances D7 and D8 are both below the predetermined threshold value, the stripe area line segments 724d and 724e are mutually orthogonal. The arrangement direction of the pixels (the χ direction and the y direction in FIG. 10) has circumscribed rectangles 727d, 727e which are parallel sides, and when at least a part of the circumscribed rectangles 727d, 727e overlap each other, the stripe area line segment 724d is permitted. And grouping, are included in the same group. On the other hand, when the distance between the line extending in one stripe region line segment and the other end point of the other stripe region line segment is longer than the threshold value, or 2 When the stripe rectangles do not overlap each other, the stripe segments are not grouped. In this way, grouping is performed to make each binary image and other binary images adjacent to the threshold. The stripe area segments overlapping each other are included in the same 312XP/invention specification (supplement)/96-08/96116068 23 1354100, and the stripe region segments that are not grouped are removed. The plurality of groups indicating the unevenness of the stripe are not obtained (step ς 18). The above-described processing by the area group obtaining unit 633 is made to be flat in the binary image adjacent to the threshold value. When the bit 2 of the stripe region of the closed region overlaps with each other, 'because the specific stripe regions are grouped and become shells, the group to be grouped is referred to as a group group in the following four days and months. In this embodiment, when a binary image is generated,

~, Tian in the soil 6 reverse thorns Peng image, use the upper limit value and the lower limit value, the shirt away from the closed value of the background value, but 疋 can also be derived from the use of the upper limit of the binary limit The derived binary image asks, the new weight p: Use the Bu Lin _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _胄#, the group obtaining unit 633 can also perform the grouping of the ... - group 7, 馀, 、, and text area segments by using the straight method. π 八 你 μ μ μ 要素 要素 要素 要素 要素 要素 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 The stripe area is a stripe unevenness element area. In the following two image areas of the target image corresponding to one stripe unevenness element area (the line is "striped uneven element area"), a stripe unevenness is required. The area display area is then 'in the stripe unevenness detecting unit 64, the display area in the target image is the target rectangular display area, the unevenness of the stripe in the target rectangle, and the adjacent rectangle 312ΧΡ/invention specification (complement) Piece)/96-08/96116068 24 1354100

The stripe unevenness element line included in each of the plurality of rectangular display areas (hereinafter collectively referred to as "adjacent rectangular area group") is determined to be the same as the stripe area updating unit 632 as shown in Fig. 8 . Specifically, when the rectangular display area 711a of the leftmost and uppermost sides of the target shirt image 7 i of FIG. 5 becomes the target rectangular display area, the object rectangular display area 7 has the right rectangular display area 711b, and the object rectangular display area has the lower right side. The rectangular display area 71L and the rectangular display area 711d below the object rectangular display area 711a become adjacent rectangular area groups. Further, the two stripe unevenness element line segments belonging to the target rectangular display area 711a and the adjacent rectangular area groups 711b to 711d are respectively specified, and the center of gravity of the two stripe unevenness element areas corresponding to the two stripe unevenness element line segments are respectively specified. The distance between the door, the σ line, and the end points of the two stripe unevenness line segments is equal to or less than the predetermined fourth threshold, and the distance between the closest endpoints of the two stripe unevenness line segments is The ratio of the length sum of the two stripe unevenness feature line segments is below the predetermined fifth threshold, and the distance between the two farthest edges of the two stripe unevenness lines and the length of the two stripe uneven feature line segments The ratio of the sum is equal to or greater than the predetermined sixth threshold. In this case, the connection of the two stripe unevenness line segments is permitted. In other cases, the link is not established (see Fig. 8). The stripe unevenness detecting unit 64 sequentially converts the target rectangular display regions in the plurality of rectangular display regions, and repeats the above-described processing (wherein the target rectangular display region may not be included in the adjacent rectangular region group), and more The stripe unevenness element line segments t have the same length direction, and the length directions are arranged close to each other, and are detected in the rectangular display areas of the adjacent 312XP/invention manual (supplement)/96-08/96116068 25 1354100 Combination of Two Stripe Uneven Feature Line Segments (Step S19) Then, the two stripe uneven line segments associated with each combination are associated with each other, and it is highly likely that the same stripe unevenness from the substrate 9 is high. The display control unit 65 derives one line segment from the two stripe uneven element line segments having the respective combinations of correlations in the same manner as described with reference to Fig. 9 . That is, the stripe unevenness component line segments are separated at the midpoint, and after the four breakpoint segments are obtained, the set of the four breakpoint segments is called the inertia main axis, and the unevenness component line segments are The end point descends to the intersection of the two intersections of the perpendicular line of the inertia main axis and the four intersection points of the inertial main axis which are the farthest from each other as the end point, and is generated as a connecting line segment connecting the two stripe unevenness line segments. Further, when one of the two stripe unevenness feature line segments has a correlation with the other stripe unevenness element line segments, the above-described processing of all the stripe unevenness element line segments is performed to generate i connected line segments. The connected line segment may be a representative line segment from the connection region (the region indicating the entire stripe unevenness) of the plurality of stripe unevenness element regions from the unevenness of the stripe (high probability). Then, the operator enters the minimum value of the length of the connected line segment to be displayed via the input unit 56. When it is accepted by the calculation unit 6 (step s2〇), as shown in the figure " j 71 ''I" value is - The length connecting line segment 731 is superimposed on the object shadow, and the sub-display is not on the display 55 (step S21). That is, the existence position of the joint region having the length of the minimum value or more is displayed on the display 55. In this manner, the operator can easily confirm only the unevenness of the stripes extending over the plurality of rectangular display regions 711 and having a minimum value or more. In addition, there is no uneven line segment of the stripe that has a connecting line segment (ie, no Ή2ΧΡ/invention specification (supplement) smelting 6-〇8/%116〇68 26 1354100 In the case where the median length is equal to or greater than the minimum value, the median length is equal to or greater than the minimum value. In addition, in the film thickness unevenness inspection dill w ^ - Λ , , -, may also be in the connection line segment Z before the display (four) 55, the plurality of stripe segments generated in each of the connected line segments belong to a plurality of regional groups In the middle of the ... The color or thickness of the segment is used to specify the unevenness of the stripe. In addition, the stripe unevenness intensity of each connecting line segment and the length of the connecting line segment (as in the manner described later) may be substantially the area of the koji region when the connecting region is obtained, respectively, and the specific intensity is The limit value and the length threshold value (area ° product limit value) can be used to determine whether the unevenness of the stripes in the set of the plurality of stripe non-sentence elements shown in the connected line segment is allowed (that is, the determination of the defect). . In the manner described above, the image processing apparatus 1 prepares a target image in which the multi-tone original image obtained by photographing the substrate 9 is shielded from the area corresponding to the gap between the rectangular display areas on the substrate 9. According to a plurality of binary images derived using a plurality of thresholds, a stripe unevenness component line segment representing a plurality of stripe unevenness element regions is specified from the target image. Then, from the plurality of stripe unevenness component line segments, two stripe uneven component line segments which are the same in the longitudinal direction and are arranged in the longitudinal direction and exist in the adjacent rectangular display region can be used to realize easy detection. A plurality of rectangular display areas extend unevenly in stripes. In this way, the 312XP/invention specification (supplement)/96·08/96116068 27 1354100 can achieve the true length or starting point or end point of the unevenness of the stripe with good precision. The result is that the stripe unevenness can be effectively specified. The cause can improve the manufacturing process or the resulting pattern. Further, in the stripe unevenness element specifying portion 63, in the plurality of stripe region line segments which are specified in the rectangular display region of the binary image, the length direction is the same, and the two are arranged in the longitudinal direction and are close to each other. The stripe area line segment is detected by the same calculation as the processing when the stripe unevenness detecting unit 64 detects the connectable stripe unevenness element line segment, and is updated to become the stripe area line segment. Here, the stripe unevenness element line segment is the longest one of the plurality of stripe region line segment segments included in the region group, and therefore the stripe unevenness element specific portion 63 performs the above-described processing on the stripe region line segment, and the stripe unevenness detection is utilized. The processing of the portion 64 is performed in the same manner, and two stripe unevenness line segments which are the same in the longitudinal direction and which are arranged in the longitudinal direction and are close to each other are detected from the plurality of stripe unevenness component line segments which are specified in one rectangular display region. The update becomes the processing of a stripe uneven feature line segment. In this manner, in the stripe unevenness element specifying portion 63, a plurality of stripe unevenness element segments having a high possibility of the same-strip unevenness in a rectangular display region are substantially updated into one stripe unevenness element. The line segment is used to effect the subsequent processing of the stripe unevenness detecting unit 或 or the display control unit 65 in the film thickness unevenness detecting device i. Further, the display control unit 65 can display the position where the connecting line segment connecting the two stripe unevenness line segments from one stripe unevenness is present on the display 55 in the connection region, and the operator can easily confirm that the plurality of rectangular display regions are covered and extended. The stripes are uneven. Next, a second embodiment of the present invention will be described. In the following description, 312 ΧΡ / invention manual (supplement) / 96 - 〇 8 / 96 丨 6 〇 68 1354100 clarification: The function of the film thickness unevenness inspection device 1 of the first embodiment is added. =12疋 block diagram, used to indicate that cpu5i in Figure 2 is used to implement CPU5 Bu_52, _3, fixed disc 54 and so on in Figure 2, indicating that the calculation unit 16 is implemented by CPU51 etc. The target image generating unit 16 includes the binary image obtaining unit 162, the stripe-specific unit 163' area group obtaining unit 164, the stripe unevenness obtaining unit 165, and the display control unit 166. In Fig. i", the stripe unevenness-specific portion 63 of Fig. 3 is simplified to the stripe-only region-departing region group obtaining portion 164, and the illustration of the stripe unevenness detecting portion 64 is omitted. It is also possible to use a dedicated circuit, and it is also possible to use a dedicated circuit in part. The procedure for checking the unevenness of the stripe by the film thickness unevenness inspection device i will be described next. Fig. 13 shows the film thickness unevenness inspection device 1 inspecting the substrate 9 The processing flow of the unevenness of the stripe on the film 92. The processing of the steps S1 to Sii8 in Fig. 13, the processing of the steps SU to S18 in Fig. 4 is the same, that is, after the plurality of binary images are captured from the target image, The specific stripe area (striped area line segment) is a stripe area line segment that overlaps each other between the binary images and the other binary images adjacent to the threshold, and is grouped in the same group. Further, the processing of step S19 in Fig. 4 is performed as needed (the illustration of the processing is omitted in Fig. 13), and the group of regions having correlation with each other in this case is used as a group of regions. The results are In the area group acquisition unit 164, the longest one of the plurality of stripe area segments included in the specific area group (hereinafter referred to as "representative strip 312XP" is obtained. / Invention Manual (Supplement)/96-08/96116068 29 1^^4100 'Line area line segment')) In addition to the line segment representing the stripe area, all the stripe area line slaves are superimposed on the representative stripe line segment. Specifically, when a specific FIG. 1=2 non-striped area line segment 724d is used as a representative stripe area line segment, each of the other strips included in the same area group as the representative stripe area line segment 724d, the text area line segment 724e is updated to be Each of the end points 725d, 726d descends to a line point 728e, 729e representing a perpendicular of the stripe area line segment 724d as a line segment of the end point.

FIG. 14 is used to illustrate the area group 73. A plurality of binary images 72 derived using the upper threshold are abstractly represented in FIG. A plurality of 2 in Fig. 14. For the hexadecimal image 72, the more the threshold value of the corresponding position is up, the length of the stripe, line & 731, 732, 733a, 733b becomes shorter as the threshold value is further away from the hall' value. In addition, as in the stripe area line segment 73 of Fig. 14, for example, 73, which is included in the same binary image and belongs to the same area group, it may be connected as needed. In the plurality of binary images, a plurality of regions representing the unevenness of the stripes are obtained, and in the stripe unevenness obtaining unit 165, the respective regions are divided into two groups: the second pass and the presence of the striped region. The image of the threshold value corresponding to the binary image corresponding to the image. For example, in the figure! Brother: Oyster's If, because the stripe area line segment stores three binary images 72, the stripe intensity of the stripe area line segment plus, such as, == coffee area group 73 becomes 3. Then, the stripe unevenness intensity of each region group and the length of the region region line segment (hereinafter referred to as "strip unevenness length") = the intensity threshold value and the length threshold value of (4) are respectively compared. Judgment 312XP/Invention Manual (Supplement)/96_Please 61] 6〇68 30 1354100 疋 Is it possible to indicate the unevenness of the stripes of the region group (that is, whether it is a defect) (Step S120). In addition, when it is determined whether or not the streak unevenness is allowed, instead of the stripe unevenness length, for example, a face of the stripe region corresponding to the representative stripe region line segment of the region group may be used, so that the area and the predetermined area are limited. The values are compared. As described above, in the film thickness unevenness inspection apparatus 1 of the present embodiment, the multi-tone target image obtained from the substrate 9 is binaryized by a plurality of threshold values, and is used to obtain respectively The plurality of binary images corresponding to the threshold value, and the stripe region of the closed region in which the flatness of each of the plurality of binary images is equal to or greater than a predetermined value. Then, in a plurality of binary shadows f, a set of stripe regions existing at substantially the same position and extending in the same direction is obtained as a group of regions indicating unevenness of the stripes, and the respective regions are acquired and striped. The range of the threshold corresponding to the binary image of the area is used as the unevenness of the stripe. In this manner, in the film thickness unevenness inspection apparatus 1, the unevenness of the unevenness of the streaks can be detected with good precision and the intensity of unevenness of the stripes can be easily obtained. In the stripe region specifying portion 163, the length direction of each closed region in the binary image is calculated, and it is specified whether or not the closed region is a stripe region, and can be used regardless of the extending direction of the closed region (the angle of the closed region) A specific stripe area with good precision. Further, a plurality of stripe region line segments in the same rectangular display region which are arranged in the longitudinal direction and which are arranged in the longitudinal direction are updated to become i stripe region line segments for unevenness from the same stripe (that is, included) Multiple stripe area segments with higher probability in the same area group are used as one strip 312XP/invention specification (supplement)/96-08/96116068 31 !3541〇〇 area line segment, which can be used in subsequent areas The group acquisition unit 64 and the stripe unevenness intensity acquisition unit 165 perform processing efficiently. ^ In the process of Fig. 13, whether or not the determination of the unevenness of the stripe on the substrate 9 is allowed is performed using the intensity threshold value related to the unevenness of the stripe unevenness, but the upper limit is determined by the operator before the above processing. The limit value is prepared in the calculation unit 16. The processing of the film thickness unevenness inspection apparatus 1 for determining the intensity threshold will be described below with reference to Fig. 15 . When the operator determines the strength threshold, similarly to the processing of FIG. 13, after the region group is obtained (step SU8), the display control unit 166 displays each of the stripe region segments to which each region group belongs. The display 55 is superimposed on the target image (step S121). At this time, as described above, except for the stripe region line segment, all the stripe region line segments are superimposed on the representative stripe region line segment, and in the display control portion 166, the stripe regions overlap each other in accordance with the respective positions on the representative stripe region line segment. The number of line segments is changed. Therefore, the line segment of the color corresponding to the unevenness of the stripe is partially or substantially completely displayed on the display 55 as the position where the group of the area is present, and the unevenness of the stripe unevenness of each of the area groups can be specified. In this way, the operator can confirm the position in the image of the object having the unevenness of the stripe together with the unevenness of the stripe, and can easily determine whether or not the intensity limit used for the unevenness of the stripe on the substrate 9 is allowed. value. The determined streak unevenness intensity is accepted by the operator using the input unit = input, and is accepted by the calculation unit 16 (step S122). Then, the unevenness of the stripe was input, and the unevenness of the plurality of substrates was examined in accordance with the above-described processing of Fig. 13. In addition, if you can specify each regional group roughly 312XP / invention manual (supplement) / 96-08/96116068 32 if strong ί, the table? For the line segment where the area group exists, the click ΤΊ β is 'any 4', for example, the degree of the line segment can also be changed. In the processing shown in FIG. 15, the imaginary line of the stripe area specifying unit 163 and the area group obtaining unit 164 is divided into six, and the processing is used to make the line segments of the line group to which the area group belongs. The threshold value has . ^ _ 八 has a corresponding relationship, in order to use the display control unit 16 6 can not show the shoes Jinzheng & ^ _ each of the temples striped unevenness intensity stripe area line ^, nA ^ Qiao Will be jealous. The processing of the 卩163 and the area group acquisition unit includes processing for obtaining the unevenness of the stripe. m Γ ", Fig. 16 is a view for explaining another example of controlling the display by the display control unit 166. In the present processing example, when the unevenness of the stripe of each region group is obtained (Fig. 13 step by step SU9) In the display 55, only the line segment indicating the presence of the area group belonging to the 条纹 庐 庐 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 预先 , , , , , It is superimposed on the pair of stalks 4 and the smock image (step S131). In this way, it is possible to specify only the necessary stripe unevenness. The article and the second time are uneven, and the other can be effectively grasped: the author confirms the region group included in the other range of the unevenness of the stripe unevenness, and the time (step S132), the operator, the face mask ^ ^ ^ ^ 轹 the author through the wheel entry portion 56 Enter the change in the set intensity range. When the calculation is performed by the calculation unit 16 scent b (step S133), the display control unit 16 changes the setting strength #瓶_夕增广β, 疋 intensity range, and only on the display 55. In the evening area group, the miscellaneous w is a line segment of the existence position of the regional group in which the intensity range is set after the change, and the weight is in the target image (the change and the representation area of the indefinite intensity range of the step author's order) 312ΧΡ/Invention Manual (Supplement)/96-08/96116068 33 1354100 The redisplay of the line segment where the group exists, repeats as needed (steps S135, S133, S134). In this way, the operator can more accurately The relationship between the actual streak unevenness and the streak unevenness intensity is grasped, and as a result, the intensity threshold used for determining whether or not the streak unevenness is allowed can be more appropriately determined. The length of the stripe unevenness (that is, the length of the line segment representing the stripe area), or the area of the stripe area corresponding to the line segment representing the stripe area of the area group, or the area of the stripe area contained in the area group 2 A representative value (for example, an average value) of a range of the threshold value corresponding to the binary image, or a line indicating the existence position of the region group may be displayed on the display 55 according to a value other than the stripe intensity, and the thickness unevenness inspection device may be used. The stripe-strength acquisition unit 165 of 1 can also obtain the sum of the intensity of the stripe unevenness in the unevenness of the stripe in each of the rectangular display regions in the target image. For example, the region group obtaining unit 1 obtains each region in which the streaks are not uneven. After the group (FIG. 13: Step S1U), (4) the stripe unevenness intensity acquisition unit 165 obtains the sum of the length of the area line segment included in each of the area groups, and 'the stripe unevenness of the area group is more A plurality of panels of 9 (adjacent parts) are respectively opposite to the shore = stripes in the rectangular display area of the image are not uneven (four) steps SU9). In this way, it is possible to obtain the degree of unevenness of the stripe of the substrate and to make it easy to determine whether or not the respective areas are allowed. One fruit, the first? A third embodiment of the invention. In the following description, the function of the film thickness unevenness inspection apparatus 1 of the embodiment is added. mxP___m/_/96 丨 34 34 1354100 The film thickness unevenness inspection device of the present embodiment is used for obtaining the formed surface of the glass substrate 9 used in the liquid-free device on one of the main surfaces. The second resist is applied for patterning: the anti-image 'checks the unevenness p on the film 92 of the substrate 9 according to the image. Here, the substrate 9 is defined as being in accordance with the second (four) boundary; H # is, usually the area of its area is more than a certain value of the brightness and darkness of the ha ha unevenness in the length of its area length = 疋 is the stripe unevenness defect, the rest are defined as the part: all: trapped usually Therefore, the unevenness of the processing on the substrate 9 may occur. = 2: If the condition can be satisfied, the strip may be appropriately changed; In addition, in the film thickness unevenness inspection installed = wrong:: light and dark fluctuations than the unevenness defect is sharp and greater than a certain degree (that is, the father, the realm of the region is relatively clear), the area is within the established range (example = in the graph In the case of the area, the diameter of the substrate 9 is below the ? surface, and the point defect of the "anomalous area in the image which occurs in the image" is a coating of the point defect * because there is a slight unnecessary object on the substrate 9, Or when the resist liquid is applied, the liquid droplets are accidentally dropped, etc. = Π is a block diagram, which is used to indicate that the CPU 51 of Fig. 2 is used to implement the CPU 51, the ROM 52, the RAM 53, the fixed optical disk 54 , etc. according to the program 541 ^ = 3 In Fig. 17, the sub-image-to-image generation unit 261 in the calculation unit 26 is realized by the CPU 51 or the like, and the streak unevenness detection unit 262, the point defect detection 3_stomach_«_/9_6116(10) 35:l\263' part is not The average detecting unit 264 masks the function of the right image of the image generating unit 265. In Fig. 17, the stripe unevenness detecting unit 262 has a binary image acquiring unit 62 and a stripe unevenness element specifying unit. Other functions can also be implemented by using a dedicated circuit or partially using a dedicated circuit.

And the owner of the person who has used the film thickness unevenness inspection device i to check the unevenness. 18A and 18B show a flow of the unevenness inspection on the film 92 of the substrate 9 by the film thickness unevenness inspection device ι. Further, Fig. 19 shows a processing flow for detecting the unevenness of the stripe by the stripe unevenness detection 4 262, and shows the processing performed in step S215 in Fig. 10 (10). The processing of step S21 and S214 in Fig. 18A and step S215 and S2154 in Fig. 19 are the same as the processing of steps SU to S18 of Fig. 4. That is, after obtaining a binary image from the target image, the specific stripe region (striped region line segment) is grouped so that each binary image and other binary images adjacent to the adjacent value are adjacent. Stripe area segments that overlap each other are included in the same group. Then, the processing of step S19 of Fig. 4 is performed as needed (in Fig. μ, the illustration of the processing is omitted), and in this case, the group of regions having correlation with each other is used as one group of regions. As a result, the stripe unevenness detecting unit 262 obtains a plurality of region groups each indicating a stripe unevenness defect. In the same manner as the representative stripe region line segment of the second embodiment, the stripe unevenness detecting unit 262 is the longest one of the plurality of stripe region line segments included in the specific region group (in the following description, it is referred to as "striping not The average area line #又"), except for the stripe uneven area line segment, all the stripe area line segment weight 312XP / invention manual (supplement) / 96 · 〇 8 / 96116068 36 1354100 璺 on the uneven area line segment. Specifically, when the stripe area line segment 724d shown in FIG. 被 is specified as the stripe unevenness area line segment, each of the other stripe area line segments 724e included in the same group group as the stripe uneven area line segment 724d is updated. The intersections 728e, 729e descending from the respective end points 725d, 726d to the perpendicular lines of the stripe unevenness line segment 724d are taken as the line segments of the end points. In the plurality of binary images 72 in FIG. 14, the more the upper limit is, the larger the threshold of the appeal is, and the length of the stripe region line segments 731, 732, 733 & The more the value leaves the background value, the shorter it becomes. Further, as the stripe area line segments 733a, 733b in Fig. 14 are included in the same binary image and belong to the same area group, they may be connected as needed. In the stripe unevenness detecting unit 262, the stripe region in the binary image corresponding to the stripe unevenness line segment is detected as the stripe uneven region indicating the streak unevenness defect (step S2155). In the present embodiment, the region of the target image corresponding to each of the stripe=average regions (hereinafter referred to as "strip unevenness region") is utilized in the present embodiment. In particular, in the case of the stripe unevenness area line segment, the processing of the above-described specific stripe unevenness area line segment is a process of detecting the stripe unevenness area. Then, an image indicating the position of the line segment of the stripe region (hereinafter referred to as "striped uneven image") is generated. In addition, instead of the uneven image of the stripe, a list indicating the position of the line segment of the uneven area may be memorized (the point defect image and the partial unevenness (four) image to be described later are also the same>. In addition, the uneven area is not used. Line-specific stripe unevenness area (ie, straight 312XP/^0J|^BJ^(^if;)/96^〇8/96 jj 6〇68 1354100 when stripe uneven area is used) except for special mention In the case where the stripe unevenness area line segment specific stripe unevenness area is used, the evaluation value acquisition unit 2661 of the determination unit 266 acquires the corresponding area group for each of the stripe unevenness area line segments. The binary image of the stripe area line segment included in the group is in the range of the threshold value at the time of binarization as the maximum intensity of the unevenness of the stripe (Fig. 18A: Step S216). For example, in the case of Fig. 14, because of the stripe area Since the line segment exists in the three binary images 72, the maximum unevenness of the stripe unevenness of the region group 73 including the stripe region line segments 73, 732, 733a, and 733b becomes 3. The maximum intensity of the stripe unevenness is made to be the stripe line. ^ has In response to the relationship, the evaluation value acquisition unit 2 6 61 records the result. Then, in the mask image generation unit 265, the stripe region line segment included in the region group in the binary image is subjected to the thickening of the predetermined pixel. Processing (expansion processing) 'In the plurality of binary images after processing, the values of the mutually corresponding pixels are added to generate a new image as shown in FIG. 2(), which is expressed in the following manner. Among the other defects, the image is substantially used to shield the image corresponding to the area of the unevenness of the stripe, so it is hereinafter referred to as "mask image" (step S217). In addition, in Fig. 2, the mouth indicates the portion of the mask f-image, and the dot line of the Xiang added (four) No. 741 indicates the stripe unevenness line segment. In fact, the area 742 of the attached line in Fig. 2 has a shading. ', Here, as described above, the binary image in the binary image - the binary value of the lower threshold or the upper threshold is detected /, in the connection = 312XP / invention manual (supplement) / 96 -08/96116068 38 1354100 A binary image with a threshold value or an upper threshold value under the background value is a closed area with a small aspect ratio at the same position, and is not limited to detecting a striped area. Therefore, when the upper threshold or the lower threshold becomes between the concentration of the uneven area in the target image and the concentration of the surrounding area of the uneven area of the stripe, 2 corresponding to the threshold The sinographic image appears in the stripe area corresponding to the uneven area of the stripe. Masking each pixel in the image, because the value corresponding to the number of binary images contained in the stripe region (correctly speaking, the stripe line segment to which the expansion process is applied) is added at the position, Therefore, the mask image substantially indicates the intensity distribution of the unevenness of the stripe unevenness in the respective uneven regions of the object image with the surrounding area as a reference. In addition, the maximum unevenness of the stripe unevenness in each stripe area is equivalent to the maximum value of the pixel in the corresponding area in the mask image. In addition, when a mask image is generated, it is also possible to use a binary image in which the line segment of the stripe region is thickened, and the stripe region is directly left in the plurality of binary images, and the binary images correspond to each other in the binary image. The values of the pixels are added to produce a mask image. When a mask image is generated, the dot defect detecting unit 263 detects a dot defect region indicating a point defect on the substrate 9 from the target image (step S218). Fig. 21 shows a processing flow in which the point defect detecting portion 263 detects the point defective region' indicates the processing performed in step S218 in Fig. 18A. In the point defect detecting unit 263, first, a value 167^ which is much larger than the background value 127 is used as a threshold value, and a pixel having a value of 167 or more in the target image is added with 1"', and "〇" is added to a pixel smaller than 167. Get J binary images. Then, with a value 87 that is much smaller than the background value of 127 as a threshold, 312XP/Invention Manual (supplement)/96-08/96116068 39

The 1JD41UU is attached to the following element plus "^"' in the figure larger than 87. The second is used to acquire another binary image (step S2181). In the case of two 2 (four) images corresponding to the two thresholds, the shrinkage processing is applied to each of the ten binary images, and the material in the graph of the graph is removed by the addition of '1'. Then, in two 2~, a logical sum of the values of the mutually corresponding pixels is generated, so that the image of the value of the pixel located is a point defect image (step S2182). In the point defect image, the value of the target image is significantly different from the background value 127 (the correct s is the pixel leaving 4G or more). The value detecting unit 263 is added. In the point defect image, the specific use mark is continuous with each other. A set of pixels of value 1 (i.e., a closed region), as shown in Fig. 22, a closed region 75 752, 753 having an area within a predetermined range is detected as a point defect region indicating a point defect (attached with τ The same symbol as the closed areas 751 to 753), the other closed areas are cut off from the point defect image (step S2183). In addition, at this stage, all of the point defect areas 751 to 753 in the point defect image do not indicate true point defects (that is, droplets are present due to the presence of minute unnecessary substances on the substrate 9 or when the resist liquid is applied). Accidental drops, etc. caused by defects. When detecting the defect area of the point, the binary image is obtained, and the value of the background value which is farther away from the average density of the target image is used as the threshold value, so that the point defect area 751 is compared with the threshold value when the uneven area is detected. ~ 753 typically has a concentration that differs from the background value by a concentration greater than the density of the streak uneven regions. Actually, the concentration of the dot defect regions 751 to 753 exceeds the variation range of the average image density of the subject image centered on the average density of the subject image (for example, in the pair 312 ΧΡ / invention manual (supplement) / 96-08/96116068 40 1354100 image The distribution of the pixel values of the image, the range of the pixels of the 5 (10)). "When the value is the center and statistically = 1 point defect area, the point defect image is selected in the shape 266: the point defect area. For example, when the point defect area: 4 in Fig. 22 is selected, the evaluation value of the determination unit 266 is obtained. The portion 2661 obtains the average value of the absolute value of the difference between the pixel value and the background value 127 as the point defect in the region corresponding to the point defect region 751 (hereinafter referred to as the defect region) in the object=image. The intensity of the point defect in the area 751 = the evaluation value A of the off. As described above, the value of the pixel of the point defect area 751 in the target image is smaller than the value of the threshold value 87 when the binary image is obtained in step (8) of FIG. 21, or is greater than the threshold value 167, so the evaluation is performed. The value A becomes 40 or more. However, the evaluation value A is obtained as the average value of the absolute value of the difference between the value of FIG. 2 included in the target image + point defect area 751 and the average density of the target image, so that the calculation of the δ level estimate A is equivalent to the representation of the target image. The average value of the values of the pixels contained in the region corresponding to the point defect P 曰 region 751 is obtained from the image of the difference between the concentration of the point defect region and the average image density of the subject image (the evaluation value E described later is also the same). Of course, the evaluation value A related to the intensity of the point defect may be a representative value other than the average value (for example, a central value) (the same applies to the evaluation values B, C, E to G described later). Then, in the mask image shown in FIG. 20, the average of the pixel values of the region corresponding to the point defect region 751 (the region indicated by the two-dot chain line with the symbol 743 attached to FIG. 2〇t) is obtained. The value is taken as the evaluation value B. Here, because the substantially all pixels contained in the area 743 in the mask image are included in the stripe 312XP/invention specification (supplement)/96-08/96116068 41 1354100, which corresponds to the stripe uneven area line segment 74丨In the rectangular region 742, the evaluation value is related to the intensity of the unevenness of the stripe shown in the stripe unevenness line segment 741 (that is, the stripe unevenness defect of the stripe uneven region corresponding to the stripe unevenness line segment 741). The assessed value. In addition, as described above, since the mask image is an image indicating the intensity distribution of the unevenness of the stripe unevenness with the surrounding area as the reference in each of the stripe regions in the target image, the evaluation value B becomes based on the target image. The value of the difference between the concentration of the uneven area of the stripe and the concentration of the surrounding area of the uneven area of the stripe. In the determination unit 266, when the evaluation value acquisition unit 2661 obtains the (four) image-like evaluation and the floor evaluation B from the mask image by the evaluation value acquisition unit 2661, the ratio of the evaluation value eight to the evaluation value B is confirmed (evaluation) Whether the value a/evaluation value B) is below the established threshold. Here, it is confirmed that the ratio of the evaluation value A to the evaluation value B of the point defect area 751 is below the threshold value, and it is determined that the point defect area 751 needs to be removed from the point defect image. Further, when the point defect area 752 in FIG. 22 is selected, the average value of the absolute values of the difference between the pixel value and the background value 127 in the point defect area 752 in the object # image is obtained as the evaluation value A, and the correlation value is obtained. The average value of the pixel values of the region corresponding to the dot defect region 752 in the mask image shown in FIG. 2A (the region indicated by the two-dot chain line to which the symbol 744 is attached in FIG. 2A) is used as the evaluation value B. Further, since the value of substantially all of the pixels included in the region 744 in the mask image is 0, the evaluation value B of the point defect region 752 becomes infinitely close to the value of 〇. Further, the determination unit 266 determines that it is not necessary to remove the pitted region 752 from the point defect image by confirming that the ratio of the evaluation value A to the evaluation value 8 of the point defect region 752 is larger than a predetermined threshold value. 312XP/Invention Manual (Supplement)/96_G8/96116068 42 1354100 In this manner, the determination unit 266 makes the density of the corresponding region in the target image and the average density of the subject image for each point defect region in the point defect image. The evaluation value A of the difference value is compared with the evaluation value B based on the value of the concentration of the corresponding region in the mask image, and is used to determine whether or not the point defect region needs to be erased (step S219). Then, the point defect area determined to be erased is cut out from the point defect image (step S22A). Actually, as in the dot defect region 752 in Fig. 22, when the dot defect region does not overlap in the stripe uneven region line segment, it is generally determined that it is not necessary to remove the dot defect region from the dot defect image. On the other hand, as shown by the point defect area 751 in FIG. 22, when the point defect area overlaps the uneven line area line #1 to 乂σ卩, the point defect area must be removed from the point defect image. The situation (not necessarily as the defect area 751 in Fig. 22 must be removed). Therefore, in step S219, the processing step of the determining portion 266 is to make the evaluation value A in the point defect region related to the intensity of the point defect when the substantially point defect region overlaps with at least a portion of the stripe uneven region line segment. And compared with the evaluation value B relating to the intensity of the unevenness of the stripe unevenness indicating the line segment unevenness area line, used to determine whether it is necessary to remove the point defect area. In addition, when the rice is considered to be removed, the point defect is considered. The area f indicates a point defect to be screened, and the point defect detecting portion 263 may become a defect to be detected area detecting portion for detecting an area indicating that the defect is to be screened. When the defect area is cut off, the determination unit 266, for example, refers to the value of the pixel in the point defect image corresponding to each of the stripe unevenness line segments in the stripe unevenness image, As shown in FIG. 23, the stripe uneven area line segment 741a is specifically included (in FIG. 23, two points 312XP / invention specification (supplement) / 96-08/96116068 43 1354100

The point defect area 754 is indicated by the chain line. Then, for the specific point defect area 754, the average value of the pixel values of the corresponding areas in the object image is obtained as the evaluation value c. Then, the maximum intensity of the streak unevenness of the stripe unevenness area line segment 741a included in the point defect area 754 is taken as the evaluation value D, and when the ratio of the evaluation value C to the evaluation value D (evaluation value C/evaluation value D) is When the threshold value is equal to or greater than the threshold value, it is determined that the stripe unevenness area line segment 741a needs to be removed from the stripe unevenness image, and when the ratio is less than the threshold value, it is not necessary to remove the stripe unevenness area line segment 74ia (step S221). At this time, since the dot defect region determined to be to be erased in the process of step S219 does not exist in the dot defect image, even if the segment of the unevenness region included in the dot defect region exists, the streak unevenness Regional line segments are also not specified. Therefore, when it is determined in step S219 that the point defect area is not required to be removed in step S219, it is checked whether or not the point defect area includes a stripe unevenness line segment, and if it is included, it is necessary to remove the stripe. Processing of uneven area segments. Then, the stripe uneven area line segment to be erased is cut out from the stripe unevenness image (step S222). When the unnecessary stripe unevenness area line segment is deleted, the evaluation value obtaining unit 2661 applies a small degree to the point defect area in the point defect image (less than the degree of expansion of the uneven portion to be screened in the step leakage described later). ) expansion treatment. In this way, the point defect areas that are close to each other are combined into i clusters (hereinafter referred to as "point defect areas" in the same manner), and the isolated point defect areas become slightly larger. Then, the average value of the value of the point defect region of the point defect area W before the expansion process in the target shirt image is shown in FIG. 312XP/invention specification (supplement)/96〇8/96116〇68 44 1354100. (In the point defect region after the expansion of the plurality of point defect regions, the average value of the pixel values in the plurality of point defect regions before expansion), as the intensity of the point defect region (hereinafter referred to as "point defect strength" ”) (step S223). In addition, in the mask image generation unit 265, the insulation value of the difference between the value of each pixel and the background value 127 is multiplied by a predetermined coefficient in the point defect region (expansion/post) in the target image, and is added. The value of the pixel corresponding to the mask image is used to update the mask image of FIG. 2 as shown in FIG. 24 (step s224). In this manner, the updated mask image becomes an intensity distribution indicating the unevenness of the stripe unevenness in the uneven area of the subject scene, and represents an image of the difference between the density of the point defect area and the average density of the subject image. Further, in the mask image of Fig. 24, the area corresponding to the point defect areas 752, 753 in Fig. 22 which is determined not to be erased by the above processing is added with a value "actually" in the areas 744, 745. According to the above-described manner, in the film thickness unevenness inspection device 1, when the dot defect defect region and at least a portion of the stripe unevenness region line segment are repeated, the determination portion 266 causes the dot defect region to be related to the intensity of the dot defect. The evaluation value can be easily judged whether or not the point defect area needs to be removed, compared with the evaluation value in the stripe uneven area line segment relating to the intensity of the unevenness of the stripe unevenness. In addition, the value of the difference between the density of the unevenness of the stripe in the target image and the concentration of the surrounding area of the uneven area of the stripe is determined by the evaluation value of the intensity of the unevenness of the stripe unevenness, even at a large aspect ratio. There are streak unevenness defects in the area, and it is also possible to use an evaluation value corresponding to the characteristic of the unevenness of the stripe, and to determine whether or not the 312XP/invention specification (supplement)/90-08/96116068 is required with good accuracy. 45 1354100 Deleting the point defect area' Then, when it is determined that the point defect area is not to be erased, check whether the point defect area contains the stripe uneven area line segment. If there is any inclusion, determine whether it is necessary to remove the uneven line area line segment. Remove unnecessary stripe uneven area segments. Next, in the thickness unevenness inspection device 1, the partial unevenness detecting portion 264 detects a portion of the unevenness to be screened portion indicating that the partial unevenness is to be screened (step S225). The partial unevenness defect herein refers to the unevenness defect except for the unevenness of the stripes among the uneven defects having a certain value or more. The detection of the partial unevenness detection area is performed by the same calculation as the detection of the point defect I1曰 area of the point defect detection unit 263, and therefore, the following is based on the figure. The processing of the partial unevenness detecting unit 264 will be explained. In the partial unevenness detecting unit 264, first, a value 132 larger than the background value 127 and around 127 is used as the threshold value, and the value of the target image is Μ? or more, and the pixel 'adds '1', which is less than 132. The pixel is appended with "〇" to get a binary image. Then, with the value 122 smaller than the background value 丨27 and around _127 as the threshold value, "1" is added to the pixel below the value of the target image, and "〇" is added to the pixel larger than 122 to obtain another One binary image (step S2181). When two binary images corresponding to the two thresholds are acquired, a contraction process is applied to each binary image, and then a small one of the sets of pixels having a value of 1 is removed by applying an expansion process. Then, in the two binary images, an image in which the logical sum of the values of the pixels corresponding to each other is used as the value of the pixel at the same position is generated as the partial uneven image to be imaged (step S2182). 312XP/Inventive Manual (supplement)/96.96116068 46 U^41〇0 2=:: Γ 影像 影像 影像 部份 部份 部份 。 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在中@上之时# 7 - Fixed value detection unit 264 'in the partial unevenness image 2 = the set of pixels whose value is not ( (that is, used;: To be:::: uneven Waiting for _ domain (step plus, please, 762, 763. If the inspection area is attached separately: "Become a defect to be screened and detected, used for:: surface defects r defects and point defects other than the defect to be screened I area In the middle of the detection of the #injury to the _ inspection area when the 'in the determination 胄%: uneven part of the image to be screened in the uneven part of the screening area. Example:: Although not selected in Figure 25 When the screening area 761 is to be inspected, the evaluation value acquisition unit 2661 obtains an area corresponding to the partial unevenness to be screened area 761 in the target image (hereinafter referred to as "partially uneven screening area") The average value of the absolute value of the difference between the value of the pixel and the background value 127 is used as the evaluation value E of the partial unevenness in the inspection area M in relation to the intensity of the partial unevenness defect. In the mask image shown in Fig. 26, the average value of the values of the pixels corresponding to the portion of the unevenness to be screened region 761 (substantially overlapping the region 771 having the shade in Fig. 26) is obtained as Evaluation value F. Here, the area in the mask image corresponding to the portion of the unevenness to be screened area 76A is substantially overlapped in the area 771 having the shade, and the area 771 is added when the mask image is updated. The area where the value of the defect area is pointed out, so the evaluation value F becomes the evaluation value related to the intensity of the point defect of the point defect area in the 312XP/invention specification (supplement)/96-08/96116068 47 1354100. When the evaluation value E of the subject image and the evaluation value F from the mask image are obtained for the partial unevenness screening area 761, the ratio of the evaluation value E to the evaluation value f is determined at the determination unit 266 (evaluation value • E/ Whether the evaluation value F) is below the established threshold. It is confirmed here. The ratio is below the threshold and is used to determine that it is necessary to remove some of the uneven screening areas from the partial uneven image to be screened. 761. In addition, when selecting part of Figure 25, the unevenness is to be sieved. In the area 762, j obtains the evaluation value E from the target image, and obtains an area corresponding to the partial unevenness to be screened area 762 in the mask image shown in FIG. 26 (the area having the shade in FIG. 26) The average value of the values of the pixels of the region 772 which is substantially overlapped is the evaluation value F. At this time, the region 772 is a region to which the value from the stripe uneven region line segment 772a shown by the one-dot chain line in Fig. 26 is added, so the evaluation value F is An evaluation value relating to the intensity of the unevenness of the stripe unevenness shown in the stripe unevenness area line segment 772a. Then, the ratio of the phase evaluation value £ and the evaluation value F by the determination unit is set to be less than or equal to a predetermined threshold value, and is used to determine that It is necessary to remove the partial inspection area 762 from the partial unevenness detection image. In addition, the aspect of the difference between the value of the pixel and the background value m in the partial unqualified screening area (10) in the image area of the image of the object to be paid is selected. The average value of the absolute values is used as the evaluation value /, and the area corresponding to the partially uneven screening area is obtained in the mask image shown in Fig. 26 (in Fig. 26, the two points of the symbol 3 are attached) The average value of the pixels of the area where the chain line is not indicated is taken as the evaluation value f. Because the mask 312XP/invention specification (supplement)/96·〇8/96116068 Δ〇 1354100, the value of the substantially whole pixel contained in the region 773 is 〇, so the partial uneven screening area 763 The evaluation value F is an infinite value close to 〇. In addition, it is confirmed that the ratio of the evaluation value E to the evaluation value F of the partial unevenness screening area 763 is greater than the predetermined threshold value, and is used to determine that it is not necessary to remove the unevenness to be screened from the image of the occupied defect image. Area 763. · In this way, in the material part 266, for the part of the uneven image to be screened, the image area corresponding to the unevenness of the image to be screened, so that (4) the object image corresponds to: the difference between the concentration of the domain and the average concentration of the object image. Evaluation of the value: The value of the concentration of the region corresponding to the image in the mask image is the same as the value of the material (step S226).鈇徭 丨 丨 丨 达 达 达 达 达 达 达 达 达 达 达 达 达 达 达 达 达 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文 文In Fig. 25, the average image to be screened is 'only a part of the unevenness to be screened. The trapped area 763 indicates that the partial unevenness is true (see Fig. 28 which will be described later). As shown in Figure 25, the unevenness of the screening area is removed. When the unevenness of the grain is to be checked, the area to be screened is not any of the areas; when the H uneven area line and the point defect area are to be sieved, it is often judged as It is not necessary to remove the portion of the unevenness to be screened area 763 from the partial unevenness. On the other side, as in Fig. 25, the unevenness of the screening area 2: =^^ When at least one part overlaps, the area of the line segment or the point defect area may be partially uneven. The part of the _mm to be screened area is not uniform. The area to be screened is 761 =: (not limited to Figure 2) - 762 is removed. Therefore, step 312XP / invention manual (supplement V96-08/ 96116068 49 Area and Article: 6: Processing 'In essence, when part of the unevenness is to be screened repeatedly, at least one part of your disc/domain line segment or point defect area is related to the estimate. #°卩不不The intensity bar of some uneven defects in the area to be screened 』===: The area line segment or the point defect area is judged. 42 The intensity-related evaluation value F is compared, and the post-processing post is used: The area to be screened. Because the step S227 field table 4 (four) is uneven (4) part of the image in the image is not uniform, the part of the image to be screened is not really partial uneven defect, = part: the same area to be screened The second = ~, the unevenness of the section (four) inspection material is called a partial uneven image. Do not part of the uneven screening area, in the judgment of 2 ί / Γ striped uneven image Each of the stripe unevenness area line segment phase - ~ Tian: the value of the pixel in the image of the unqualified image, as shown in Figure 27: the chain contains the uneven portion of the line segment (in Figure 27: : chain) The uneven area m. Then, the mean value of the pixel of the region corresponding to the specific value W764 in the target image is obtained as the evaluation value G. Then 'the uneven portion of the region is m kg=the stripe uneven region line segment 741b's stripe unevenness maximum intensity ^evaluation H' When the ratio of the evaluation value G to the evaluation value H (evaluation value g/evaluation=H) is determined to be a stripe uneven image stripe uneven region at a predetermined threshold value In the line segment 741b, when the ratio is less than the threshold value, the stripe unevenness area line segment 74113 is not removed (step S228). At this time, it is determined by the processing of step S226 that it is necessary to remove the uneven portion to be screened because It exists in the partial unevenness (to be screened) image 3i2XP/invention manual (supplement)/96-08/96116068 50 1354100, so even if there is a stripe uneven area line segment included in the portion of the unchecked area to be screened , the stripe uneven area line segment is also not specified. Therefore, step S228 t The process of the determination unit 266 determines that it is not to be removed in step S226. When the P-unevenness is to be screened, it is confirmed whether the portion of the unevenness is to be screened, whether or not the line segment is unevenly distributed, and if it is included, It is determined whether or not the stripe unevenness line segment is to be removed. In addition, the specific portion also includes a partial uneven region of the point defect region, and the uneven region is obtained in the region corresponding to the object image. The average value of the pixel value is used as the evaluation value G. Then, the point defect intensity of the point defect area included in the partial uneven area is used as the evaluation value H, and when the ratio of the evaluation value G to the evaluation value 在 is established When the value is equal to or greater than the limit value, it is determined that it is necessary to remove the dot defect region from the dot-abstracted shirt image, and when the ratio is less than the threshold value, the dot defect region is not to be erased (step sm). Then, the strip that needs to be removed, "the uneven area line segment or the point defect area is removed from the uneven image or the point defect image (step S229). When Tian Fancheng does not need the defect area and the stripe uneven area line segment is removed" The partial unevenness of the partial uneven image is applied to a large extent (greater than the degree of expansion of the defective region of the point S223). In the above manner, the adjacent portions of the uneven regions are combined. The clusters (hereinafter referred to as "partially uneven regions"), the area of the uneven portion 763 formed in Fig. 28 is increased in size (wherein, the unevenness after the expansion processing is indicated by a broken line in Fig. 28 The screening area 763 & indicates the partial unevenness of the screening area 1 which is removed in step S227 by the second = chain line. Then, for each partial uneven area 763a, the object image 312XP / invention manual (repair) is obtained. Face 嶋 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 Average of the values of the pixels contained ) as the part: S23 ^ degree (hereinafter referred to as "partial unevenness defect strength") (step in the film thickness unevenness inspection tw - device 1, according to the need for the uneven area of the stripe uneven area and part For each of the uneven areas, the intensity obtained is also 7 I". The intensity of the unevenness is large, the point defect strength or the partial unevenness defect and the area of the area (in the case of uneven unevenness of the stripe, the stripe = domain line segment The specific area of the uneven area of the stripe or the length of the stripe unevenness line segment, and the specific f-degree threshold and the area threshold (length threshold) of the defect type are respectively compared with each other. Whether or not to allow the indication of defects in the area. In addition, for the stripe unevenness domain segment 'point defect area and part of the uneven area, only the specific 2 degree range belongs to, or the specific area range (length range) belongs to There is a position, for example, a colored line or the like corresponding to its area or the like may be used, and a specific intensity or area may be displayed in the display device 55. In this case, the operator can confirm that it is displayed on the display 55. In the trapped area, the intensity threshold area threshold for determining whether or not to allow is input via the input unit 56. As described above, the film thickness is in the unevenness inspection device, and the unevenness of the portion to be screened and the unevenness of the stripes are uneven. At least part of the weight of the area line segment or the point defect area, the evaluation unit 266 uses the evaluation value relating to the intensity of the partial unevenness defect of the partial area, and the unevenness of the stripe 312^P/invention specification ( Supplement) /96-08/96116068 52 1354100 The evaluation value of the intensity of the unevenness of the stripe unevenness in the area line segment, or the evaluation value related to the strength of the point defect of the point defect area. In this way, it is easy to It is determined whether it is necessary to remove a part of the uneven screening area, and as a result, the stripe uneven area line segment, the point defect area, and the partial unevenness (to be screened) area can be correctly detected. Here, if the plurality of partial unevenness screening areas 761 to 763 in FIG. 25 are not all removed, when the expansion processing is applied in step S23, the dotted line is indicated by the dotted line of the additional symbol 736b in FIG. The portion of the average defect or the point defect that is not to be subjected to the screening area 761 becomes included in the area indicating the last part of the unevenness defect or the point defect. On the other hand, in the film thickness unevenness inspection apparatus of the present embodiment, the strength of the stripe unevenness defect or the point defect is larger than the intensity of the unevenness of the adjacent unevenness in the nails of the plurality of portions to be unchecked. It is used to achieve an area indicating partial unevenness with good precision. Further, in the determining unit 266, when it is determined whether or not the partial uneven screening target area is to be removed, the evaluation value related to the partially uneven screening area is substantially obtained in the middle of the object image. The image of the difference between the concentration of the uneven screening area and the average wave width of the soil image is the representative value of the value of the pixel contained in the area corresponding to the portion of the uneven screening product. The evaluation value related to the uneven area of the disc stripe or the point defect area, the intensity distribution of the unevenness of the stripe in the uneven area of the strip representing the current image of the object = the difference between the density of the defect area and the average density of the target image Uneven (4) The value of the pixel contained in the area corresponding to the inspection area. In this way, 'determine whether to remove some of the uneven screening 312XP/invention manual (supplement)/96•嶋u UM100 = indicates the concentration of partial uneven defects and the unevenness of the streaks or the corresponding fields The area and the partial unevenness of the screening area ^ θ / derived 2 of the 5 estimates, can be used to appropriately (4) in addition to the partial unevenness of the inspection area. In the other film thickness unevenness inspection device 1, when it is judged that the portion U is not to be cut (four), it is confirmed that the portion of the unevenness to be screened includes i, the uneven line segment or the point defect region, If there is a case, it is determined whether or not the stripe uneven area line segment or the point defect area is to be removed, and it is possible to appropriately remove the unnecessary stripe uneven area line segment. In this way, it is possible to obtain the intensity of each defect by precisely marking the uneven line segment, the point defect area, and the partial defect. As a result, it is possible to determine whether or not the defect is allowed with the accuracy of the defect. Further, when it is considered that the occurrence of the difference n is effective, it is possible to effectively form the stripe unevenness defect, the point defect and the partial unevenness defect, and the manufacturing process or the formed pattern can be improved. Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and various modifications are possible. In the first embodiment, the film thickness unevenness inspection device 1 may perform the streak unevenness inspection without using the stripe unevenness element line-specific stripe unevenness element region in the stripe unevenness element specifying portion 63. In the case of the ten-month situation, when the stripe area unevenness area is specified, the stripe area updating unit 632 has the same length direction in each binary image, is arranged in the length direction, and is updated in a plurality of stripe areas close to each other. In the area group acquisition unit 633, the area group acquisition unit 633 has two striped areas adjacent to each other in the 2 312XP/invention specification (supplement)/96-08/96116068 54 1354100 image. Grouping into the same group "used to find the area of the stripe unevenness element, which is the area of the stripe unevenness area, because it is included in the fringe area of the area group to make the length direction Since the length of the stripe area updating unit 632 is substantially the same in the plurality of stripe unevenness element areas which are characteristic of a rectangular display area of the binary image, the detection length direction is the same and is arranged. The two stripe unevenness element regions that are close to each other in the longitudinal direction correspond to the process of updating the two stripe unevenness element regions into one stripe unevenness element region. In addition, the processing performed by the unevenness detection detection unit 6 is to detect that the lengths of the plurality of stripe unevenness element regions are the same, and are arranged in the longitudinal direction, and are present in two adjacent rectangular display regions. In the stripe unevenness element region, the display control unit 65' generates a connection region in which the two stripe unevenness element regions detected are connected, and the display 55 displays the existence position in the connected region by the line indicating the edge of the connection region. Make it overlap the object image. However, in the case where it is easier to detect the stripe extending over a plurality of rectangular display areas, it is preferable that the stripe unevenness element specific portion 63, and the stripe unevenness element area table (4) of the stripe area have a money section (four). In the stripe unevenness detecting unit 64 and the display control unit 65, the stripe unevenness element line segment is used as the stripe unevenness element region corresponding thereto, thereby simplifying the processing. Two on = real, in the form is to use a plurality of thresholds from the object image, and the plurality of 2 (four) image-specific stripes that are derived are not uniform (or the stripe uneven feature area, but can also be used for the image) In the binary image, the length of the 312XIV invention specification (supplement)/96-08/96116068 55 1354100 has a length and a ratio of the width perpendicular to the longitudinal direction that is a predetermined value or more, directly as a plurality of closed regions. In the first embodiment, the stripe unevenness detecting unit 64 obtains each combination of two stripe unevenness element line segments existing in the adjacent rectangular display region, and passes through 4 2 The distance between the center of gravity of the stripe unevenness feature line segment and each end point, and the distance between the nearest end point and the distance between the most distal point between the two stripe unevenness line segments are obtained. (4) Two stripe uneven feature line segments The ratio of the length sum is used to detect the two stripe uneven feature line segments that can be connected, but the two stripe unevenness feature line segments that are determined to be connectable are arranged in the length direction if they are at least the same in the length direction, and respectively In the case of two rectangular display areas adjacent to each other, it is not necessarily necessary to be close to each other. The detection of the two stripe unevenness element line segments that can be connected may be performed by other calculations, for example, for existing adjacent ones. The various combinations of the two stripe unevenness element line segments in the rectangular display area extend the two stripe unevenness element line segments, and the angle formed by the two extended straight lines is obtained. In the first embodiment, the substrate 9 is formed. A film of a resist is formed thereon, but if it is formed by applying a liquid on the main surface, the material of the film formed on the substrate may be a material other than the resist liquid. The film thickness unevenness inspection device 1 is particularly suitable The inspection of the unevenness of the stripes on the glass substrate, but it is also possible to use a plurality of rectangular display regions having the same pattern to be arranged in an empty gap, and the stripes on the other substrates in which the plurality of rectangular display regions are multi-angled are not In the film thickness unevenness inspection device 1 of the second embodiment, the stripe region 312XP/invention specification (supplement)/96-08/96116068 56 1354100 specific portion 16 It is also possible to perform the stripe unevenness inspection as the stripe area line segment. In this case, the stripe area specifying section 163 updates the length direction of the same 'the plurality of stripe areas that are arranged in the longitudinal direction and are close to each other'. In the 1st solid stripe area, the area group obtaining unit 164 groups the stripe areas that are closest to each other in the binary image adjacent to the threshold to be included in the same area group, and respectively The area group of the unevenness of the stripes is obtained by the stripe unevenness obtaining unit 165: the threshold corresponding to the binary image of the striped area is present in each of the area groups, or the stripe area is in the longitudinal direction. The sum of the lengths and the areas is the unevenness of the stripe. The display control unit 166 displays the overlap in the target image, for example, in the plurality of stripe regions included in the group of regions, and the longest direction in the two directions indicates the edge of the representative stripe region. Line, you can confirm the strip = the area where the unevenness exists. However, it is easy to detect the unevenness of the streaks and to obtain the unevenness of the streak. It is preferable to specify the stripe area in the stripe region specifying portion 163 by using the representative line segment, and obtain it in the group of regions; The acquisition unit 165 can simplify the processing by using the stripe area corresponding to the 'kilogram area' in the stripe area line segment. In the case of the group 164, there is a case where only one part of the group is obtained. The acquisition unit 165... 1, in the case of the use area group = in the case of the film thickness unevenness inspection device, at least 1 is obtained by using the material. = group=unevenness intensity acquisition unit 165 obtains the positional unevenness intensity of each of the target and the at least one of the target regions, and displays the presence position of the target image to the magical region group by the display control unit (10), or At least _ invention manual (suggested detail: 5, 1354100 the existence of a specific one in a regional group.

In the second embodiment, it is not necessary to have a value that is inevitably continuous in the target gift 9A, and it is not necessary to use both of the upper threshold values when the plurality of thresholds are deducted. The number of 2nd generation generated by the binary image acquisition unit 162 may be 2 or more, and may be an arbitrary number. The calculation of the amount of calculation of the nasal portion 16 is reduced by -1: the unevenness of the stripe is obtained. In view of the above, it is preferable to make the == L limit value to obtain a binary image of 5 or more. =, the interference light from the substrate 9 is received in the photographic section 41 to obtain the original shirt image, but the intensity of the interference light is periodic; the original image is changed when the thickness of the film on the substrate to be inspected is changed. The degree (the average value of the pixel values) also changes. In the above-mentioned implementation 2 2: in the image generation unit 16 or the target image generation unit, 'J is described by the multi-pattern (1) to (4). The above-mentioned = average density is a substantially constant target image, and when the ratio of the target image 162 (or the stripe unevenness detecting unit 262, the dot defect = the partial unevenness detecting unit 264, the same applies hereinafter) The m threshold value 1 is such that the interference of the interference light is not used when the image is captured, or the threshold value of the binary image acquisition unit 16 = 2 is changed to match the average density of the original image. The original image obtained by the photographing unit is directly targeted; a plurality of binary images are obtained from the image of the second image (or the uneven area of the stripe, the point defect area, and the partially uneven area to be screened) can be directly detected. That is, the hex image is obtained 胄162, which becomes the object of the binary object = or 2 3 ΠΧΡ / invention manual (supplement) / 96-08/96116068 58 1354100 Detecting uneven stripes 'point defect area and part The image is not the same as the original image or the image derived from the original image. In addition, in the stripe region specifying portion 163, it is not necessary to obtain a torque for the 2 = closed region, and for the closed image == The smallest product has a length in the longitudinal direction and a ratio perpendicular to the length of the long sound = direction, which is compared with a predetermined value to determine whether the closed region is a striped region. However, when a plurality of stripe regions extending radially from a certain position are formed at various angles (for example, when a film of a resist is formed on a substrate by a coating device using a spin coating method, there is an unnecessary presence on the substrate before the treatment. The object is caused by the unnecessary object as the starting point to form the radial stripe I. When the stripe area is specified with good precision, it is preferable to use the moment of the closed area to calculate the stripe area in order to reduce the amount of calculation. . In the film thickness unevenness inspection apparatus 1 of the second and third embodiments, the substrate to be inspected is not necessarily formed as a film such as a resist film. In addition, the film thickness unevenness inspection apparatus 1 is particularly suitable for use for inspection of unevenness of stripes on a substrate which is intended to be cut at a plurality of locations, such as a glass substrate or a semiconductor substrate of a display device, but an object other than the substrate may be used. Check the unevenness of the stripes. In the unevenness inspection device 1 of the third embodiment, when the point defect region or the partial unevenness to be screened region overlaps with at least a portion of the stripe unevenness region (or the stripe unevenness region line segment), the determination point Defect area or part of the uneven screening area to be removed, but compared with the uneven area of the stripe to determine 312 χ ρ / invention manual (supplement) / 96-08/96116068 59 1354100 疋 whether to eliminate the defect, it can also be Indicates an area that is somewhat defective or partially uninjured or that is to be screened, or that indicates that there are defects other than those to be screened. In other words, in the film thickness unevenness inspection device 1, when the unevenness of the unevenness of the table = streak unevenness is detected, the target image 2 is detected by the image, and when there is another defect in the image to be sieved, When the defect-to-screening area is overlapped with at least a portion of the uneven area of the stripe, the evaluation value related to the defect strength of the defect-to-screening area and the evaluation of the stripe unevenness defect intensity in the stripe=average area are evaluated. When the values are compared, it is used to determine whether or not the defective area to be screened is to be removed, and as a result, the uneven area of the stripe and the area to be screened for the defect can be surely detected. In addition, similarly, it is determined whether the defect is to be removed compared with the point defect area, and may be an area indicating that both the unevenness of the stripe defect and the partial unevenness defect are to be screened, or may indicate that there is a defect other than the defect. The area to be screened. In other words, in the film thickness unevenness inspection device ,, when the dot defect region indicating the dot defect is detected, the object image is binarized for performing the detection “When there is another defect in the object image to be screened. When the defect to be screened is overlapped with at least a portion of the point defect area, the evaluation value related to the defect strength of the defect to be screened area is compared with the evaluation value related to the point defect intensity of the point defect area, It is used to easily determine whether the defect to be screened area is to be removed, and as a result, the point defect area and the defect to be screened area can be correctly detected. In step S221 of Fig. 18A of the third embodiment, it is determined whether or not the stripe unevenness line included in the point defect area is to be removed, but it is also possible to determine whether or not the stripe uneven area line segment is large 312XP. /Inventive manual (supplement)/96-08/96116068 60 1354100 = The length of the line segment with respect to the unevenness of the stripe (or the length of the stripe uneven area/a length (area) having a certain ratio or more close to 1) is included = , trapped in the area. Similarly, in step S228 of Fig. 18β, it is judged whether or not the stripe uneven area line segment or the point defect area to be erased may be included in the partial uneven (four) (four) domain. In addition, in the third embodiment, the stripe unevenness detecting unit== binary image is used to group the stripe regions in each binary image, and (4) detecting the stripes of the target scene. In the uneven area, the uneven area of the stripe is used to indicate that the ratio of the length in the length direction to the width in the direction of the direction is not less than the predetermined value. (4): The detection of the uneven area of the stripe can also be performed by other methods. == Bu, in the detection of uneven area of the stripe, point defect area and partial unevenness (4) 适 适 地 地 更 更 更 更 更 更 更 更 更 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 艮 另外 另外 另外 另外However, in the case of good: accuracy detection of defect areas and partial unevenness of the area to be screened, j' preferably uses multiple thresholds to binarize the object image for more than 4 The image is obtained by finding the logic of the pixel values corresponding to each other in the binary image and the 'wire detection point defect area and part ^: area 2: another 'the same method can be used to detect the uneven area of the stripe' In this case, the stripe area specified by the upper threshold and the stripe area specified by the lower threshold are used as the same type. The invention is described, but the description above is for illustrative purposes only and is not intended to limit the invention. Therefore, many variations or aspects may be made without departing from the scope of the invention. 312XP/Invention Manual (Supplement)/96-〇8/96Π 6068 61 丄 [Simple description of the drawing] Figure 1 shows the structure of the thickness unevenness inspection device. Figure 2 shows the construction of a computer. Figure 3 is a block diagram ' used to represent the functional configuration implemented by the computer. β 4 is a flow of the process of inspecting the unevenness of the streaks on the substrate by the film thickness unevenness inspection device of the first embodiment. Fig. 5 shows an object image. Figure 6A shows a binary image. Figure 6A shows a binary image. Fig. 6C shows a binary image. Figure 7 shows the closed area. Fig. 8 is a diagram for explaining the determination of whether or not the stripe region line segments are connected. Figure 9 is used to illustrate the connection of the line segments in the stripe region. Fig. 10 shows that the stripe region line segments in the two binary images are superimposed. Fig. 11 shows that the connected line segment is superimposed on the target image. Figure 12 is a block diagram ' used to indicate the functional configuration implemented by the computer. Fig. 13 is a flow chart showing the process of inspecting the unevenness of the stripes on the substrate by the film thickness unevenness inspection device of the second embodiment. Figure 14 is used to illustrate the group of regions. Fig. 15 is a view showing the flow of another process of the film thickness unevenness inspection device. Fig. 16 is a view showing the flow of still another processing of the film thickness unevenness inspection device. Figure 17 is a block diagram showing the functional configuration implemented by a computer. Fig. 18 is a flow chart showing the processing of the unevenness defect on the substrate 312ΧΡ/invention specification (supplement)/96-〇8/96116068 62 1 in the film thickness unevenness inspection device of the third embodiment. Fig. 18 is a flow chart showing the processing of the unevenness defect in the film thickness unevenness inspection in the third embodiment. Fig. 19 shows a processing flow for detecting a stripe unevenness region. Figure 20 shows a mask image. Fig. 21 shows the processing flow of the detection point defect area. Fig. 22 shows a point defect area.

Checking the device to inspect the substrate Fig. 23 shows the point defect area. Figure 24 shows the updated mask image. Figure 25 shows a partially uneven image to be screened. Fig. 26 shows a mask image. Fig. 27 shows a partial uneven area. Fig. 28 shows a partial uneven area. Fig. 29 shows a partial uneven area of the comparative example. [Description of main element symbols] 1 Film thickness unevenness inspection device 2 Stage 3 Light irradiation unit 4 Light receiving unit 6 and 26 Calculation unit 8 Recording medium 9 Substrate 16 Calculation unit 312ΧΡ/Invention manual (supplement)/96-08/ 96116068 63 1354100 21 Moving mechanism 31 Halogen lamp 32 Quartz rod 33 Cylindrical lens 41 Photographic unit 42 Lens 43 Filter 51 CPU 52 ROM 53 RAM 54 Fixed disc 55 Display 56 Input 56a Keyboard 56b Mouse 57 Reading device 58 Communication unit 6 261 target image generating unit 62, 162 binary image acquiring unit 63 stripe unevenness element specifying unit 64, 262 stripe unevenness detecting unit 65 display function control unit 71 target image 72 binary image 312XP/invention manual (supplement /96-08/96116068 64 1354100 73 Area group 91 Main surface 92 Resist film 161 Target image generation unit 165 Stripe unevenness acquisition unit 166 Display control unit 211 Motor 212 Guide 263 Point defect detection unit 264 Part unevenness detecting unit 265 mask image generating unit 266 determining unit 541 programs 631 and 163 stripe region specifying unit 63 2 Stripe area update unit 633, 164 Area group acquisition unit 2661 Evaluation value acquisition unit 312XP/Invention manual (supplement)/96-08/96116068 65

Claims (1)

1354100 X. Patent application scope: The domain is set to be arranged in a vertical and horizontal direction with empty gaps, and the unevenness of the stripes on the substrate with multiple de-angles is checked. 1·:: Film thickness unevenness inspection method 'Multiple rectangles having the same pattern The area is in accordance with the plurality of rectangular areas described above; and is characterized by comprising: (a) the step of photographing the substrate, masking the area of the gap, and preparing the target image; (b) in the binary image obtained by binarizing the domain material, the length of the length direction and the perpendicular direction are The ratio of the width in the longitudinal direction is equal to or greater than a predetermined value, (4) the step of defining a plurality of closed regions as the plurality of stripe regions; and (C) detecting the same length direction from among the plurality of stripe uneven element regions. The two stripe uneven element regions are arranged in the longitudinal direction and exist in the adjacent rectangular regions. 2. In the method of inspecting the film thickness unevenness of the patent specification 帛i, in the step (b), the uneven pattern element region is represented by the stripe unevenness line segment representing the region, in the above step (4) The middle stripe unevenness element line segment is used as the corresponding stripe unevenness element area. 3. The method for inspecting the film thickness unevenness according to the ninth application of the patent application, further comprising: (b) using the same calculation as the above step (c), detecting a plurality of specific ones in a rectangular region The stripe unevenness is in the money domain, and the length direction is the same, and is arranged in the longitudinal direction, and the two adjacent stripe unevenness element regions are updated to become one stripe unevenness element region. A step of. 312XP / invention manual (supplement) / 96_ 〇 8 fine 丨 6 〇 68 66 1354100 4. If you apply for patent scope! The film thickness unevenness inspection method, = = film on one main surface (four). Top: =1 Encloses the first item - check (bl) uses multiple thresholds to make the above objects separately matched with the above multiple threshold handles for the retrieval; 祁 corresponding multiple binary The step (b2) of the image is a step of specifying a stripe region in which the ratio of the width in the longitudinal direction = the length direction is a predetermined value or more in each of the plurality of binary shadows, and (10) at the threshold value. In the adjacent binary image, the overlapping stripe regions are grouped and included in the same region group to obtain respectively: Read = at least one region group, and (4) specifying the plurality of stripes unevenly Step of the element region; the method for inspecting the film thickness unevenness further includes: (4) obtaining, within each of the at least one region group, a threshold range corresponding to a binary image in which a striped region exists, or a stripe region The sum of length or area as a step of unevenness of streaks. 6. The film thickness unevenness inspection method of claim 5, wherein the method further comprises: (e) making the average stripe unevenness intensity of each of the at least one region group specific, and The step of presenting at least one of the area groups is displayed on the display unit. 7. For the film thickness unevenness inspection method according to item 5 of the patent application scope, 3 UXP/invention manual (supplement)/96-08/96116〇68 67 1354100 further includes: (f) The step of displaying the existence position of the range of the specific stripe unevenness intensity among the at least one of the plurality of area groups is displayed. 8. The film thickness unevenness inspection method according to Item 7 of the patent application, comprising: (f) the step of changing the intensity range of the unevenness of the stripe, and displaying the at least one of the at least one region group in the display portion The step of changing the position of the stripe unevenness intensity after the change belongs to the position of the owner. 9. The film thickness unevenness inspection method of claim 5, wherein in the step (b2), a stripe region representing a region of the region is used to specify a stripe region, in the above (1) 3) step and the above ( In the step d), the stripe region line segment is used as the stripe region corresponding thereto. 10. The film thickness unevenness inspection method according to the fifth aspect of the invention, wherein in the step (b2), calculating a moment of a closed region in each of the plurality of binary images, determining a length direction, thereby specifying Whether the above closed area is a striped area. 11. The film thickness unevenness inspection method according to item 5 of the patent application, wherein in the step (b2), the plurality of stripe regions having the same length direction and arranged in the longitudinal direction and being adjacent to each other are updated into a stripe region. 12. The film thickness unevenness inspection method according to claim 5, wherein the total of the plurality of rectangular region unevenness intensities is obtained in the step (4). 13. The film thickness unevenness inspection method according to item 1 of the patent application scope, wherein: 312ΧΡ/invention specification (supplement)/96.〇8/96116〇68^1354100 by performing steps (a) to (c) above From the above, the unevenness of the stripe unevenness region indicating the unevenness of the stripe is detected. (4) The width of the average area in the length direction is not longer than the predetermined value; the width is perpendicular to the length direction. The method for inspecting the film thickness unevenness further includes: a step of detecting a defect to be screened, wherein the image of the object is (g) a binarized region of the object image, and the defect to be screened area table His defects are to be screened; and (8) When the defect-to-screening area is overlapped with at least the above-mentioned stripe unevenness area, the defect strength of the above-mentioned defect to-be-detected area is evaluated, and the remaining unevenness intensity of the above-mentioned stripe uneven area is compared. The second (fourth) value is used to determine whether or not to remove the defect to be screened. 14. The method for checking the thickness unevenness according to item 13 of the patent application, further comprising: 〃 φ (丨) detecting an area from the object image within a predetermined range, and having an average concentration with the image of the object a region in which the difference is larger than the concentration of the uneven portion of the stripe as a point defect region indicating a point defect; (j) when the defect to be screened and at least a portion of the point defect region are repeated The third evaluation value of the defect strength of the defect-to-screening area described above is compared with the fourth evaluation value of the point defect intensity of the above-mentioned point defect area, and is used to determine whether or not the defect-to-screening area is to be removed. 15. For the film thickness unevenness inspection method according to item 13 of the patent application, 69 312XP/invention specification (supplement)/96-〇8/%116〇68 The above first evaluation value is based on the above-mentioned inspection area The concentration and the above object are both; 1: the second evaluation value is the value of the difference between the above objects, the difference between the indifferences of the surrounding areas of the above η, and the film thickness of the 15th item The unevenness inspection method, wherein the first evaluation value is an average concentration of the target image, the concentration of the area to be screened, and the area corresponding to the inspection area, and the image is not The representative value of the intensity of the unevenness of the streaks in the average area, the area of the area corresponding to the defect to be screened, and the method of the film thickness unevenness inspection method of the fifteenth item of the patent application, in the step, When it is determined that it is not necessary to remove the above-mentioned defect to be _, 'confirm whether or not the above-mentioned defect to be screened region contains the above-mentioned stripe is not large +, and in the case of inclusion, it is determined whether or not the above-mentioned stripe/uneven area is to be erased. The film thickness unevenness inspection method according to any one of the items 1 to 17, further comprising: (k) displaying, on the display unit, a position where a connection region connecting the two stripe unevenness element regions exists . 19. The film thickness unevenness inspection method of claim 18, wherein 'more includes: 312XP/»_ book (supplement)/96·Ο8/96116068 70 1354100 (1) in the above step (k) Previously, the minimum value of the input length of the connected region indicating the existence position in the above step (k) is accepted. 20. A film thickness unevenness inspection device, wherein a plurality of rectangular regions having the same pattern are arranged to be vertically and horizontally arranged with an empty gap, and unevenness of streaks on a substrate having a plurality of chamfers is inspected according to the plurality of rectangular regions; The photographing unit is configured to capture the original image of the multi-tone image by using the substrate, and the target image generating unit uses a region corresponding to the gap of the original image as a mask to generate a target image; In (4) the fixed part, in the image obtained by binarizing the image of the upper riding, the length in the longitudinal direction and the ratio perpendicular to the length of the length and the direction are in a plurality of closed areas above the shouting value, And the unevenness detecting portion; and the unevenness detecting portion is configured to measure the same length direction from the plurality of stripe uneven element regions, and the two stripes are unevenly arranged in the adjacent rectangular region The symplectic area is located in the 不 不 不 申请 如 如 如 如 如 = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = t; used in the uneven pattern inspection area. & as the corresponding unevenness of the stripe element 22. The film thickness unevenness inspection device according to claim 20, in the above-mentioned stripe unevenness-receiving portion, And the upper line pattern unevenness inspection 312XP / invention said 0 month book (supplement) / 96, lion 61 丨 6 〇 68 71 1354100 test. P processing the same calculus, detection is specified in a rectangular area In the stripe unevenness element region, the same length direction is arranged in the longitudinal direction, and two stripe unevenness element regions are adjacent to each other, and the two stripe uneven element regions are updated to become one stripe unevenness element region. ' 23. The film thickness unevenness inspection device of claim 20, wherein the film formed on the main surface of the film is uneven. The substrate has a liquid application layer. 24. The inspection device according to the second to third aspects of the patent application, further comprising a display portion for displaying a position where the connection region connecting the two stripe unevenness element regions exists. 25. The film thickness unevenness inspection device according to claim 24, wherein the minimum value of the input length of the connection region indicating the position on the display portion is received before the display portion is displayed. 26. A computer-readable recording medium in which a program is recorded. The computer sets a plurality of rectangular areas having the same pattern to be arranged in an empty space and vertically and horizontally, and checks the stripes on the corners according to the plurality of rectangular areas. It is characterized in that it is executed on the computer when executed by the above computer: 34 @ (a) For the substrate photographing, the area corresponding to the obtained multi-tone original image gap is masked, and the object image is prepared. +^ (b) In the second step of the above-mentioned object image, the 312XP/invention specification (supplement)/96-08/96116068 72 1354100 in the hard image makes the length in the length direction perpendicular to the above. a ratio of the width in the longitudinal direction to a predetermined value or more, and a plurality of closed regions as a plurality of stripe uneven element regions; and (c) detecting the length direction from among the plurality of stripe uneven element regions Similarly, the steps are arranged in the above-described longitudinal direction and exist in two stripe uneven element regions in the adjacent rectangular regions. 27. For the recording medium of claim 26, In the step (b), the stripe unevenness element region is represented by the stripe unevenness element line segment representing the region, and the stripe unevenness element line segment is used as the corresponding stripe unevenness element region in the step (c). use. 28. The recording medium of claim 100, wherein the method further includes: (b) the step of detecting the plurality of stripe unevenness elements specified in a rectangular area by using the same calculation as the above step (c) In the region, the two stripe unevenness element regions are arranged in the same length direction and arranged in the longitudinal direction, and the two stripe unevenness element regions are updated to form one stripe unevenness element region. 29. The recording medium of claim 26, wherein the substrate has a film formed by coating a liquid on one main surface. The recording medium according to any one of claims 26 to 29, wherein when the program is executed by the computer, the computer is further executed: (d) displaying the two uneven stripes on the display unit The step of the location of the connected area of the feature area. 312XP/Inventive Manual (Repair)/96-08/96116068 73 1354100 31. The recording medium of claim 30, wherein when the above computer is used to execute the above program, the computer is further executed: (e) in the above (d) Before the step, the step of inputting the minimum value of the input length of the connected region in which the existing position is displayed in the above step (d) is accepted. 312XP/Invention Manual (supplement)/96_08/96116068 74