An IC photolithography character recognition and detection device surface imaging system and method

Abstract

The application discloses an IC photoetching character recognition and detection device surface imaging system and method, which comprises a "claw" type light source clamp, an arc-shaped light source clamp, a camera clamping mechanism, a gantry fixing mechanism and a conveying mechanism. The imaging system comprises a bar-shaped light source, a ring-shaped light source, a backlight light source, a telecentric lens and a CCD industrial camera. The method comprises chip positioning, pin positioning and image data processing. The IC photoetching character recognition and detection device surface imaging system and method comprise a light source clamp, a camera clamp and a chip positioning recognition method and implementation steps. The light source adjustment and camera angle adjustment need to be combined with corresponding visual algorithms according to actual operation conditions to realize more accurate, faster and more suitable chip surface photoetching character recognition and further realize chip defect detection and classification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor visual inspection, in particular to an IC photolithography character recognition and detection device surface imaging system and method. BACKGROUND

[0002] The use of chips is almost in all electronic devices, and the character information on the chip is important information for marking the product. Therefore, it is crucial to distinguish the characters on the chip; therefore, the demand for fast character recognition technology applied in industrial production is particularly obvious; especially the visual design scheme of the visual inspection device and the algorithm process. In the actual production process of semiconductor chips, the measurement of the pin size and the overall chip size of the chip and the detection of the character mark on the chip surface are generally included; the detection of the character includes identifying the content of the character, the quality of the character printing, and whether the character is missing or not. Based on the demand for character recognition detection, the character position needs to be located to make further decisions; based on the detection of defects and the measurement of size, more effective image features of the chip need to be characterized.

[0003] Patent CN110097048A provides a SOT chip image rapid correction and character recognition method, which can be used for rapid identification and detection of SOT chips.

[0004] Patent CN111274961A provides a flexible IC substrate character recognition and information analysis method, which can be used for detection of variable length strings and identification of environmental factors.

[0005] Patent CN113221889A provides a chip character anti-interference recognition method and device, which can effectively reduce part of the interference of the environment when collecting images, prevent the chip from being out of angle, and improve the production efficiency.

[0006] Patent CN109685070A provides an image preprocessing method, which solves the technical problems of low accuracy of character detection in the IC sealing process in the prior art.

[0007] Patent CN110852328A provides a chip back character recognition method, which has the characteristics of being able to remove a large amount of noise of the photographed picture and not needing to correct the chip.

[0008] Patent CN111330862A provides a chip double positioning and adjusting device, which can detect whether there is a problem on the upper and lower surfaces of the chip at the same time.

[0009] Patent CN111754461A provides a positioning method and device for the character region of a semiconductor chip image; by proposing an angle point filtering method and improving the convex hull detection algorithm, the character region of the semiconductor chip image can be more accurately positioned.

[0010] The patent CN107767373A provides a chip pin fast sub-pixel precision visual detection method, which realizes fast and high-precision detection of chip pin width.

[0011] The recognition of chip surface characters will inevitably have some character defects, such as character loss, character printing not obvious, chip surface scratches affecting character recognition, etc. Some of these existing defects can be reduced by optical design to reduce the difficulty of visual algorithm design, and further easily classify the type of defects and make corresponding visual algorithm decisions. At the same time, each part of the chip has certain characteristics that can be represented under different conditions, so a light source with a specific angle needs to be designed for illumination to achieve the effect of specific feature information being represented. For example, chip pins, chip characters, chip sides, and chip overall surface illumination; different positions of illumination need to consider the combination of the entire illumination system without affecting each other, while cooperating with the design of visual algorithm, to adapt to the applicability of different chips in different environments.

[0012] The patent CN111366588A provides a semiconductor chip character surface defect pin defect detection light source, which solves the problem of three light sources in the same track superposition of the old version, and is compatible with other product detection.

[0013] The patent CN213516936U provides a detection mechanism of an IC chip detection device, which can adjust the corresponding mechanism to realize the illumination of the light source on the characters or pins, so as to achieve better detection effect.

[0014] The patent CN210629661U provides an OCR character acquisition device based on deep example vision, in which the shooting angle of the camera and the light source and the placement position of the character carrier can be quickly adjusted to achieve the effect of quickly and accurately acquiring characters.

[0015] The patent CN104580860A provides an auxiliary device for reading chip surface characters, which realizes the preprocessing work of chip image acquisition, and enables the chip to quickly and stably perform image acquisition.

[0016] The patent CN104881665A provides a chip character recognition and verification method and device, which reduces the influence of character breakage and adhesion on character recognition.

[0017] The patent CN110927184A provides a visual detection method for chip packaging defects, which realizes the effects of good imaging, accurate positioning, and high detection efficiency.

[0018] The patent CN108982508A provides a kind of based on feature template matching and deep learning's plastic package IC chip defect detection method, solve the chip positioning and character detection, and the problem of detecting and classifying chip pin defect, meet certain IC chip's online detection needs.

[0019] Based on the description of the above various patents, basically, a large number of specific chip recognition and defect classification problems are solved.But the function implementation of each patent has certain applicability of specific chip.For the detection and recognition device of semiconductor chip system and the algorithm simple design application of system, especially the detection system involving multiple camera operations, multiple light source light path design and visual algorithm mutual cooperation, there is no chip detection visual system that can specifically adapt to multiple environments on the market.

[0020] For the needs of chip character recognition and defect detection, chip pin feature representation and measurement, etc.The application provides an IC photolithography character recognition and detection device, surface imaging system and method. SUMMARY

[0021] The purpose of the present application is to provide an IC photolithography character recognition and detection device surface imaging system and method to solve the problem of low efficiency of detecting a type of chip on multiple devices, manual judgment of whether the chip character is correct or not, and lack of multi-azimuth angle adjustment of traditional light source fixture and camera fixture.

[0022] To achieve the above purpose, the present application provides the following technical scheme: an IC photolithography character recognition and detection device surface imaging system, the IC photolithography character recognition and detection device includes a "claw" type light source fixture vertically placed, an arc-shaped light source fixture placed on the side, a camera clamping mechanism with multi-azimuth adjustment, a gantry support frame and a conveying mechanism.

[0023] In the surface imaging system, it includes a bar light source, a low-angle ring light source, a backlight light source, a telecentric lens and a CCD industrial camera.

[0024] Among them, the related light source is controlled according to the characteristics of the light source, that is, the characteristics of different features can be controlled by coding whether the light source illuminates the IC sample and the brightness of the illumination, to avoid waste of light source and possible impact.

[0025] Preferably, the vertically placed "claw" type light source clamp first end is connected with the gantry support frame, and the vertically placed "claw" type light source clamp and the gantry support frame are connected through a rotating fixed block to form a left-right rotating adjustment structure, the rotating fixed block is connected with the "claw" type light source fixing frame through a connecting rod structure of the "U" shaped connecting block, the first end of the "U" shaped connecting block is connected with the "convex" shaped connecting block integrated with the side end of the rotating fixed block, and the vertically placed "claw" type light source clamp can form an inclination angle adjustment structure through the "convex" shaped connecting block, the end of the "U" shaped connecting block is connected with the "claw" type light source fixing frame, and the "claw" type light source fixing frame forms a further telescopic action adjustment structure, the "claw" type light source fixing frame is uniformly distributed with hole structures for mounting various strip-shaped light sources and ring-shaped light sources, and the strip-shaped light sources and ring-shaped light sources are used for illuminating the IC sample.

[0026] By adopting the technical scheme, the overall mechanism has positions for fixing light sources in various directions, and can connect corresponding light sources through pins and has a certain space position adjustment function. Through the adjustment function of the light source clamp, the light source can reach a more ideal illumination angle in space. The light source clamp adjustment mainly includes two steps, one is preliminary adjustment, which has left-right, inclination and telescopic adjustment functions, and the other is accurate adjustment, which is the further fine adjustment between the light source and the "claw" type fixing frame through pin connection.

[0027] Preferably, the side-placed arc-shaped light source clamp body is an arc-shaped honeycomb plate, the arc-shaped honeycomb plate as a whole has an arc shape, and the arc-shaped honeycomb plate has honeycomb-shaped fixing holes for fixing corresponding backlight light sources; the arc-shaped honeycomb plate is connected with the gantry support frame, and the arc-shaped honeycomb plate forms an inclination angle adjustment structure; the backlight light source and the hole in the honeycomb plate are connected through a "U" shaped connecting rod to form an angle adjustment structure; the arc-shaped honeycomb plates are symmetrically installed on both sides of the conveying mechanism to achieve symmetrical illumination effect on the IC sample.

[0028] By adopting the technical scheme, the side-placed arc-shaped light source clamp body has an arc shape, and the arc-shaped plate has honeycomb-shaped fixing holes for fixing corresponding light sources. The arc-shaped plate is connected with the gantry frame to adjust the inclination angle, and further, the light source and the hole in the honeycomb plate are connected through a pin to finely adjust the angle. The adjustment steps of the side-placed arc-shaped light source clamp mainly include two steps, one is inclination adjustment, and the other is fine adjustment of the honeycomb plate connected with the pin.

[0029] Preferably, the end of the camera clamping mechanism with multi-directional adjustment is fixedly provided with a CCD industrial camera, and the mechanism at the end of the camera clamping mechanism with multi-directional adjustment mainly comprises three rectangular honeycomb fixing plates, which mainly serve to fix the position of the CCD industrial camera, and the rectangular honeycomb fixing plates can fix multiple cameras, and the CCD industrial camera is connected with the rectangular honeycomb fixing plates through a camera fixing connecting block, the camera fixing connecting block constitutes a plane rotation angle adjusting structure, the rectangular honeycomb fixing plates constitute a rotating structure around the middle camera fixing connecting block to realize the multi-directional angle adjustment of the camera, the first end of the rectangular honeycomb fixing plates is connected with a rack connecting guide block, the camera clamping mechanism with multi-directional adjustment constitutes a pitch angle adjusting structure at the front end of the rack connecting guide block, and the end of the rack connecting guide block is fixedly connected with a rack structure, the rack structure fixed at the end of the rack connecting guide block is a guide rail structure connected with a rectangular protection plate, and the rack structure fixed at the end of the rack connecting guide block adjusts the forward and backward movement of the camera clamp end through the stroke control of the rack and pinion adjusting knob, and the whole camera clamping mechanism with multi-directional adjustment constitutes a left-right adjusting structure at the gantry support frame.

[0030] By adopting the technical scheme, the end of the camera clamping mechanism with multi-directional adjustment mainly comprises three "honeycomb-shaped" fixing plates, and the honeycomb holes are mainly used to fix the position of the camera. Further, the forward and backward movement of the camera clamp end can be controlled by adjusting and controlling the rack and pinion knob. The camera clamp adjustment mainly comprises two steps, one is preliminary adjustment, which has the functions of left-right, pitch and forward-backward adjustment, and the other is accurate adjustment, which is the further fine adjustment between the CCD industrial camera and the honeycomb-shaped fixing plate through a pin.

[0031] An IC photoetching character recognition and detection device surface imaging method, the imaging method comprising chip overall positioning, pin positioning and image data processing flow;

[0032] The chip overall positioning method mainly highlights the overall characteristics of the chip and removes unnecessary noise through the cooperation of light source imaging, extracts the bright area by using a binary algorithm, further calculates the geometric information of the bright area, and corrects the chip placement position by affine transformation processing. Further, the selection of the light source includes a low-angle ring light source, a bar light source and a backlight source;

[0033] The pin positioning method scans the chip pins in the form of a scanning line to determine the specific placement position of the chip pins. Further, a positioning method combining template matching and feature analysis is proposed, which can further correct the angle of the required transformation to the correct position to realize the chip positioning and recognition function;

[0034] The image data processing flow method, the image pre-processing in the early stage mainly includes transforming the image on the color space, so as to obtain the average data of all images; by setting a threshold range, the pre-processed image effect map to be selected is determined.

[0035] Preferably, in the IC optical character recognition process, in order to accurately and quickly recognize characters, there are certain requirements for the posture of the recognized characters, so the algorithm is further designed to correct the character posture. In the rectangular image, some characters in the rectangular image may be in an inverted posture, and a positioning algorithm is further designed to ensure that the characters are in an upright state.

[0036] Based on the feature analysis calculation of the chip, there are two methods provided: one is to filter and count the number of chip pins to determine whether the current situation of the chip is in an upright posture. The second is to form a horizontal scanning line on the rectangular image, and scan from top to bottom, and intersect with the pin features; the intersection result is used to determine whether the current situation of the chip is in an upright posture. Among them, the initial position formed by the scanning line and the working interval of the scanning line are the keys to ensure the accuracy of scanning and improve the running speed; the specific threshold needs to refer to the calculation result of the actual chip position.

[0037] Preferably, in the method of pin positioning of the horizontal scanning line, the pin positioning of the chip is mainly judged by forming a scanning line to scan the chip pins. Further explanation, the initial position formed by the scanning line is to set an initial coordinate threshold by calculating the center position of the characters on the chip surface, and further set an interval threshold between the scanning lines. The main functions of the threshold setting are: first, the threshold setting can avoid unnecessary scanning line formation, thereby saving the consumption of computer memory and further saving time consumption; second, the interval setting between the scanning lines can effectively avoid the wrong judgment caused by the different lengths of the chip pins, and also saves the scanning time of the computer and further improves the running speed of the algorithm. Third, in the scanning process, the specific position of the chip pin can be judged by setting the threshold, further the function of pin measurement can be realized, further the chip pin defect detection function can be realized according to the front and back comparison of the intersection result of the scanning line and the pin.

[0038] Preferably, the chip pin defect detection function is to mark the corresponding result when the corresponding image pre-processing is performed, so as to realize the effect of defect classification. For example:

[0039] First, in selecting the best processing channel, the distribution of the brightest pixel of the character is calculated, and the result can be used to determine whether the character is printed thickly or thinly. If the character is printed thickly, the value of the brightest pixel is higher, and vice versa.

[0040] Second, when counting the connected domain of the character, the current statistical result can be used to determine whether the character printed on the current chip has the phenomenon of character loss.

[0041] Third, before obtaining the overall connected domain of the character, the entire character needs to be processed by difference, and the difference between the standard character and the character to be recognized is calculated. The result of the difference can be used to determine whether the current character is a few strokes, many strokes, or broken.

[0042] In the statistical classification of defect classification, all the corresponding algorithm marks are made in the necessary image preprocessing to make decision classification, and large-scale neural network training is not needed, so that a better and faster classification and recognition effect is achieved.

[0043] Compared with the prior art, the IC lithography character recognition and detection device, the surface imaging system and the method have the beneficial effects that,

[0044] 1. By adjusting the light source clamp and the CCD industrial camera clamp, the light source and the camera can present multiple illumination and shooting postures in space, can meet the operation in multiple complex environments in actual industrial operation, and multiple implementation methods and process steps are given, so that a chip detection vision system that can adapt to multiple environments is provided.

[0045] 2. The problem that a type of chip needs to be detected on multiple devices and manually distinguished whether the chip character is correct or not is solved, and the problem that the traditional light source clamp and the camera clamp lack multi-azimuth angle adjustment is solved. DETAILED DESCRIPTION

[0046] Figure 1 It is a schematic view of the vertical "claw" type light source clamp structure of the application.

[0047] Figure 2 It is a schematic view of the arc-shaped light source clamp structure placed on the side of the application.

[0048] Figure 3 It is a schematic view of the camera clamping mechanism structure with multi-azimuth adjustment of the application.

[0049] Figure 4 It is a general flowchart of the IC character recognition method of the application.

[0050] Figure 5 It is a template matching and edge line angle adjustment flowchart of the application.

[0051] Figure 6 For the annular light source lighting schematic of the present application;

[0052] Figure 7 For the character normalization process flowchart of the present application;

[0053] Figure 8 For the defect character classification system schematic of the present application;

[0054] Figure 9 For the scanning line algorithm flowchart of the present application;

[0055] Figure 10 For the feature analysis positioning recognition algorithm flowchart of the present application;

[0056] Figure 11 For the overall structure schematic of the present application.

[0057] In the figure: 1, vertically placed "claw" type light source clamp; 101, rotating fixed block; 102, "convex" type connecting block; 103, "U" type connecting block; 104, "claw" type light source fixing frame; 2, side placed arc light source clamp; 201, "U" type connecting rod; 202, arc honeycomb plate; 3, camera clamping mechanism with multi-directional adjustment; 301, gear rack adjustment knob; 302, rack connecting guide block; 303, camera fixed connecting block; 304, rectangular honeycomb fixed plate; 4, gantry support frame; 5, conveying mechanism; 6, strip light source; 7, annular light source; 8, backlight light source; 9, telecentric lens; 10, CCD industrial camera; 11, IC sample. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0059] Please refer to Figures 1-11 The present application provides a technical solution: an IC lithography character recognition and detection device surface imaging system, which comprises a vertically placed "claw" type light source clamp 1, a side placed arc light source clamp 2, a camera clamping mechanism with multi-directional adjustment 3, a gantry support frame 4, and a conveying mechanism 5.

[0060] In the surface imaging system, there are a strip light source 6, a low-angle annular light source 7, a backlight light source 8, a telecentric lens 9, and a CCD industrial camera 10.

[0061] The related light source is controlled according to the characteristics of the requirements, that is, the different characteristics need to be characterized to control the light source through coding to control whether the light source illuminates the IC sample 11 and the brightness of the illumination, so as to avoid waste of the light source and possible influence.

[0062] The vertical "claw" type light source clamp 1 is connected to the gantry support frame 4 at the first end, and the vertical "claw" type light source clamp 1 and the gantry support frame 4 are connected through the rotating fixed block 101 to form a left and right rotating adjustment structure, the rotating fixed block 101 is connected to the "U" type connecting block 103 through the connecting rod structure of the "U" type connecting block 103, and the "U" type connecting block 103 is connected to the rotating fixed block 101 at the side end of the "convex" type connecting block 102, and the vertical "claw" type light source clamp 1 can form an adjustment structure of the pitch angle through the "convex" type connecting block 102, and the "U" type connecting block 103 is connected to the "claw" type light source fixing frame 104 at the end, and the "claw" type light source fixing frame 104 forms a further telescopic action adjustment structure, and the "claw" type light source fixing frame 104 is uniformly distributed with hole structures for the installation of various strip light sources 6 and ring light sources 7, and the strip light sources 6 and the ring light sources 7 are used for the illumination of the IC sample 11, and the schematic diagram of the vertical "claw" type light source clamp 1 is shown in Figure 1 .

[0063] The overall mechanism has a fixed light source position in each direction, and the corresponding light source can be connected through a pin, and has a certain space position adjustment function. Through the adjustment function of the light source clamp, the light source can reach a more ideal illumination angle in space. The light source clamp adjustment mainly includes two steps, one is the preliminary adjustment, which has left and right, pitch and telescopic adjustment functions; the other is the precise adjustment, which is further fine adjustment between the light source and the "claw" type fixing frame through the pin connection.

[0064] The arc-shaped light source clamp 2 is placed on the side, and the main body of the arc-shaped light source clamp 2 is an arc-shaped honeycomb plate 202, and the arc-shaped honeycomb plate 202 has an arc-shaped shape as a whole, and the arc-shaped honeycomb plate 202 has a honeycomb-shaped fixing hole for fixing the corresponding backlight light source 8; the arc-shaped honeycomb plate 202 is connected to the gantry support frame 4, and the arc-shaped honeycomb plate 202 forms a pitch angle adjustment structure; the backlight light source 8 and the hole on the honeycomb plate are connected through the "U" type connecting rod 201 to form an angle adjustment structure; the arc-shaped honeycomb plate 202 is symmetrically installed on both sides of the conveying mechanism 5 to play a symmetrical illumination effect on the IC sample 11, and the schematic diagram of the arc-shaped light source clamp 2 placed on the side is shown in Figure 2 .

[0065] The arc-shaped light source clamp 2 placed on the side presents an arc-shaped form, wherein the arc-shaped plate has honeycomb-shaped fixing holes to fix the corresponding light source. The arc-shaped plate is connected with the gantry to adjust the angle of inclination. Further, the light source is connected with the hole on the honeycomb plate through a pin to make fine angle adjustment. The adjustment steps of the arc-shaped light source clamp placed on the side mainly include two steps, i.e. inclination adjustment and fine adjustment of the honeycomb plate connected with the pin.

[0066] The CCD industrial camera 10 is fixedly installed at the end of the camera clamping mechanism 3 with multi-directional adjustment, and the mechanism at the end of the camera clamping mechanism 3 with multi-directional adjustment mainly consists of three rectangular honeycomb fixing plates 304. The rectangular honeycomb fixing plate 304 mainly serves to fix the position of the CCD industrial camera 10. The rectangular honeycomb fixing plate 304 can fix multiple cameras, and the CCD industrial camera 10 is connected with the rectangular honeycomb fixing plate 304 through a camera fixing connecting block 303. The camera fixing connecting block 303 constitutes a plane rotation angle adjustment structure. The rectangular honeycomb fixing plate 304 constitutes a rotation structure around the middle camera fixing connecting block 303 to realize spatial multi-directional angle adjustment of the camera. The rectangular honeycomb fixing plate 304 is connected with a rack connecting guide block 302 at the first end. The camera clamping mechanism 3 with multi-directional adjustment constitutes a tilt angle adjustment structure at the front end of the rack connecting guide block 302. The rack connecting guide block 302 is fixedly connected with a rack structure at the end. The rack structure fixed at the end of the rack connecting guide block 302 is a guide rail structure connected with a rectangular protection plate. The rack structure fixed at the end of the rack connecting guide block 302 controls the forward and backward movement of the camera clamp end through the rack and pinion adjustment knob 301 to adjust the stroke of the rack and pinion. The camera clamping mechanism 3 with multi-directional adjustment constitutes a left-right adjustment structure at the gantry support frame 4. The schematic diagram of the camera clamping mechanism 3 with multi-directional adjustment is shown in Figure 3 .

[0067] The end of the camera clamping mechanism 3 with multi-directional adjustment mainly consists of three "honeycomb-shaped" fixing plates. The honeycomb holes are mainly used to fix the position of the camera. Further, the rack and pinion knob can be adjusted to control the forward and backward movement of the camera clamp end. The camera clamp adjustment mainly includes two steps. The first step is preliminary adjustment, which has left-right, tilt and forward-backward adjustment functions. The second step is fine adjustment, which is the further fine adjustment between the CCD industrial camera and the honeycomb-shaped fixing plate through a pin.

[0068] An IC photoetching character recognition and detection device surface imaging method, the imaging method includes chip positioning, pin positioning and image data processing flow;

[0069] The positioning method of the whole chip is mainly through the cooperation of the light source to highlight the whole chip features and remove unnecessary noise, and the binary algorithm is used to extract the bright area, and the geometric information of the bright area is further calculated, and the affine transformation is processed to correct the chip placement position. Further, the selection of the light source includes a bar light source 6, a low-angle ring light source 7 and a backlight light source 8;

[0070] The positioning method of the pin is to scan the chip pin in the form of a scanning line to determine the specific placement position of the chip pin. Further, the positioning method combining template matching and feature analysis is proposed, which can further correct the angle of the required transformation to the correct position to realize the chip positioning and recognition function;

[0071] The image data processing flow method, the early image preprocessing mainly includes transforming the image in the color space to obtain the average data of all images; by setting a threshold range, the preprocessed image effect diagram to be selected is determined.

[0072] The image data processing flow method, the images collected by the imaging system may not fully meet the design of the visual algorithm, so it is necessary to further optimize the quality of the collected images. The early image preprocessing mainly includes transforming the image in the color space to obtain the average data of all images; a threshold range is set to determine the preprocessed image effect diagram to be selected.

[0073] The image obtained from the camera is split into R, G, B three-channel images, and further converted into H, S, V three-channel images. The sum of all gray values at each coordinate point is calculated using the six images calculated, and the result is the average value of the gray value. A threshold is set to determine the best processing channel, and finally the output is a single-channel image. The specific gray value average calculation formula is described as follows, assuming that the gray values of each channel image are represented as f R (x, y), f G (x, y), f B (x, y), f H (x, y), f S (x, y), f V (x, y) Then take the average of the gray values of each channel, and finally add the average values of each channel to obtain the single-channel image gray value, that is,

[0074]

[0075] In the formula, n represents the number of channels of the current input image, and ux,y is the output image. In practical application of the formula, the formula in the numerator is not necessarily the input image listed, but is appropriately removed or added according to the actual situation. In general, the mode of the formula is used for calculation.

[0076] Further, if the average gray value calculated in the formula does not reach the set threshold range, the image channel is reselected and the above steps are repeated until the desired optimal processing channel is output.

[0077] The determined optimal image processing channel can be further calculated to form a minimum circumscribed rectangle with a direction; the angle and position of the minimum circumscribed rectangle are calculated, and the rectangle region of the character ROI region is obtained by translation and rotation. At this point, the preliminary positioning has obtained the position of the overall chip.

[0078] In order to further correct the placement position of the chip, a shape-based template matching algorithm is also proposed to further determine the position and angle of the character feature. Because when calculating the angle and position coordinates of the highlighted feature, some extreme positions may occur, which may cause position correction errors and further lead to recognition errors. For example, if the placement position of the IC is exactly at the critical angle position of the calculation, it may not be able to further obtain the correct character placement position, but a position that is 90° different from the expected result. The critical position may appear when the angle between the middle line of the IC upper and lower pins and the vertical center line is 80°-90°, or the angle between the middle line of the pins and the horizontal center line is 0°-10°. The specific angle correction method and steps are as follows: first, obtain the rectangular image of the preliminary positioning, then calculate the edge angle θ1 of the character rectangular region in the middle of the chip, second, obtain the template angle θ2 found by template matching, and obtain the angle θ by subtracting the obtained angles from each other. θ is used to further determine the correct state of the current chip. If the value of θ is not in the predetermined threshold range, the above operation is repeated until the desired value is obtained. The specific implementation flowchart is shown in Figure 5

[0079] ​In the visual imaging system module, the lighting of the light source is essential to highlight the target features. For the light source color, the angle and distance of the light source irradiation, etc. The choice may cause certain imaging difficulties to the imaging system. The preferred light source of the present light source is a white ring-shaped light source, the irradiation angle of the lamp bead is 0°-15° with the sample, and the distance between the light source and the lens is between-150mm-250mm. Among them, the definition of distance is based on the center point of the object as the particle. For example, the lens is the horizontal and vertical center intersection as the particle, and the light source is similar. The distance between the two is established as the coordinate system with the two ends coinciding as the boundary line. Moving up is positive, and moving down is negative direction. The angle involved is explained with the horizontal plane as 0°. Figure 6

[0080] In the process of character feature segmentation, the connected domain of the character may have certain adhesion or rupture. It is necessary to normalize the character to realize the integrity and uniformity of the character. The implementation method and steps are as follows: a minimum circumscribed rectangle is formed by obtaining a character area, a segmentation threshold is set according to the width and height of the character, a relatively independent area of each character is obtained, and an intersection is further made between the character and the corresponding pixel position of the original character area, and finally a complete segmented character area is obtained. The specific algorithm flow is shown in the following Figure 7

[0081] In the process of IC lithography character recognition, in order to accurately and quickly recognize the character, the character posture has certain requirements, so the algorithm is further designed to correct the character posture. In the rectangular image, some characters in the rectangular image may be in an inverted posture, and a positioning algorithm is further designed to ensure that the character is in a normal posture.

[0082] Based on the feature analysis calculation of the chip, there are two methods: one is to filter and count the number of chip pins to determine whether the current situation of the chip is in a normal posture. The second is to form a horizontal scanning line on the rectangular image, and scan from top to bottom, and intersect with the pin feature. The intersection result is used to determine whether the current situation of the chip is in a normal posture. Among them, the initial position formed by the scanning line and the working interval of the scanning line are the key to ensure the accuracy of scanning and improve the running speed; the specific threshold needs to refer to the calculation result of the actual chip position, and the specific algorithm flow of the two positioning methods is shown in the following Figure 10

[0083] ​​​The method for pin positioning by horizontal scanning line is mainly to scan the chip pins in the form of scanning line to determine the specific placement position of the chip pins. Further, the initial position of the scanning line is set by calculating the center position of the characters on the chip surface to set an initial coordinate threshold, and further set an interval threshold between the scanning lines. The main role of the threshold setting is: first, the threshold setting can avoid unnecessary scanning line formation, thereby saving the consumption of computer memory and further saving time consumption; second, the interval setting between the scanning lines can effectively avoid the wrong judgment caused by the different lengths of the chip pins, and also saves time for computer scanning, further improving the running speed of the algorithm. Third, in the scanning process, the specific position of the chip pins can be determined by setting the threshold, further realizing the function of pin measurement, and further judging whether the chip pins exist the lack of pins according to the front and back comparison of the intersection result of the scanning line and the pin, realizing the defect detection function of the chip pins. The specific scanning algorithm process is as follows Figure 9 as shown.

[0084] The defect detection function of the chip pins is to mark the corresponding results when corresponding image preprocessing is performed to realize the effect of defect classification. For example:

[0085] First, when selecting the best processing channel, the distribution of the brightest pixels of the characters is calculated, and the calculated result can be used to determine whether the characters on the surface are printed full or light. If the characters are printed full, the value of the brightest pixels calculated is higher, otherwise, the result value is lower.

[0086] Second, when counting the connected domain of the characters, the current statistical result can be used to determine whether the characters printed on the current chip exist the phenomenon of character loss.

[0087] Third, before obtaining the overall connected domain of the characters, the entire character needs to be processed by difference, and the difference between the standard character and the character to be recognized is processed by difference. The result of the difference can determine whether the current character is a few strokes, many strokes or broken, etc.

[0088] In the statistical classification of defect classification, all the corresponding algorithm marks are made in the necessary image preprocessing to make decision classification, and do not need to perform large-scale neural network training, so as to achieve good and fast classification and recognition effect. Some specific defect chip classification charts are shown below Figure 8 as shown.

[0089] What is not described in detail in the specification is the prior art known to those skilled in the art, although embodiments of the application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the application, the scope of the application is defined by the appended claims and their equivalents.

Claims

1. An IC photolithographic character recognition and inspection device surface imaging system, characterized by: The IC photocharacter recognition and detection device comprises a vertically placed "claw" type light source clamp (1), a laterally placed arc-shaped light source clamp (2), a camera clamping mechanism (3) with multi-directional adjustment, a gantry support frame (4), and a conveying mechanism (5); In the surface imaging system, a bar light source (6), a low-angle ring light source (7), a backlight light source (8), a telecentric lens (9), and a CCD industrial camera (10) are included; The related light source controls the illumination of the light source according to the characteristics represented by the requirements, that is, different characteristics are represented by coding to control whether the light source illuminates the IC sample (11) and the brightness of the illumination; The vertically placed "claw" type light source clamp (1) is connected to the gantry support frame (4) at the first end, and the vertically placed "claw" type light source clamp (1) and the gantry support frame (4) are connected through a rotating fixed block (101) to form a left-right rotating adjustment structure; The rotating fixed block (101) is connected to the "claw" type light source fixing frame (104) through the connecting rod structure of the "U" type connecting block (103), and the "U" type connecting block (103) is connected to the "convex" type connecting block (102) integrated on the side end of the rotating fixed block (101), and the vertically placed "claw" type light source clamp (1) can form an elevation angle adjustment structure through the "convex" type connecting block (102); The "U" type connecting block (103) is connected to the "claw" type light source fixing frame (104) at the end, and the "claw" type light source fixing frame (104) forms a further telescopic action adjustment structure, and the "claw" type light source fixing frame (104) is uniformly distributed with hole structures for mounting various bar light sources (6) and ring light sources (7), and the bar light sources (6) and ring light sources (7) are used for illuminating the IC sample (11); The laterally placed arc-shaped light source clamp (2) has an arc-shaped honeycomb plate (202), and the arc-shaped honeycomb plate (202) as a whole presents an arc-shaped shape, and the arc-shaped honeycomb plate (202) has a honeycomb-shaped fixing hole for fixing the corresponding backlight light source (8); The arc-shaped honeycomb plate (202) is connected to the gantry support frame (4), and the arc-shaped honeycomb plate (202) forms an elevation angle adjustment structure; The backlight light source (8) is connected to the hole in the honeycomb plate through a "U" type connecting rod (201) to form an angle adjustment structure; The arc-shaped honeycomb plate (202) is symmetrically installed on both sides of the conveying mechanism (5), and has a symmetric illumination effect on the IC sample (11). The camera clamping mechanism (3) with multi-directional adjustment is fixedly installed with a CCD industrial camera (10) at the end, and the mechanism at the end of the camera clamping mechanism (3) with multi-directional adjustment is specifically composed of three rectangular honeycomb fixed plates (304), which specifically function to fix the position of the CCD industrial camera (10), and the rectangular honeycomb fixed plate (304) can fix multiple cameras, and the CCD industrial camera (10) and the rectangular honeycomb fixed plate (304) are connected through a camera fixed connecting block (303), and the camera fixed connecting block (303) constitutes a plane rotation angle adjustment structure; The rectangular honeycomb fixed plate (304) constitutes a rotating structure around the middle camera fixed connecting block (303) to realize the spatial multi-directional angle adjustment of the camera; The rectangular honeycomb fixed plate (304) is connected with a rack connecting guide block (302) at the first end, and the camera clamping mechanism (3) with multi-directional adjustment constitutes a pitch angle adjustment structure at the front end of the rack connecting guide block (302), and the rack connecting guide block (302) is fixedly connected with a rack structure at the end; The rack connecting guide block (302) is fixedly provided with a rack structure at the end, which is connected with a rectangular protection plate to form a guide rail structure, and the rack structure fixedly connected with the rack connecting guide block (302) at the end adjusts the stroke of the rack and pinion through a rack and pinion adjustment knob (301) to control the forward and backward movement of the end of the camera clamp, and the camera clamping mechanism (3) with multi-directional adjustment constitutes a left-right adjustment structure as a whole at the gantry support frame (4).

2. An IC photolithographic character recognition and detection device surface imaging method, characterized by: The imaging method includes the positioning of the whole chip, the positioning of the pins and the image data processing flow; The positioning method of the whole chip highlights the features of the whole chip and removes unnecessary noise through light source cooperation and illumination imaging, extracts bright areas by using a binary algorithm, further calculates the geometric information of the bright areas, and corrects the chip placement position by affine transformation processing, and the selection of the light source includes a low-angle ring light source, a strip light source and a backlight light source; The positioning method of the pins scans the chip pins in the form of a scanning line to determine the specific placement position of the chip pins, and proposes a positioning method combining template matching and feature analysis, which can further correct the angle of the required transformation to the correct position to realize the chip positioning and recognition function; The image preprocessing method in the early stage specifically includes transforming the image in the color space to obtain the average data of all images, and determining the preprocessed image effect diagram to be selected by setting a threshold range.

3. The method of claim 2, wherein: In order to accurately and quickly recognize characters, the recognized character posture is required during the IC lithography character recognition process, so the algorithm is further designed to correct the character posture, and in the corrected rectangular image, some characters in the rectangular image are in an inverted posture, and the positioning algorithm is further designed to ensure that the character is in a vertical state. Based on the feature analysis and calculation of the chip, there are two methods provided: First, by screening and counting the number of chip pins, it is determined whether the current situation of the chip is a vertical posture. Secondly, on the image of the rectangular area, a horizontal scanning line is formed, and the scanning is from top to bottom, intersecting with the pin features; whether the current situation of the chip is a normal posture is judged by the intersection result; The initial position formed by the scanning line and the working interval of the scanning line are the keys to ensure the accuracy of scanning and improve the running speed, and the specific threshold needs to be referred to the calculation result of the actual chip position.

4. The method of claim 3, wherein: In the method for positioning the pin by the horizontal scanning line, the chip pin is scanned by forming a scanning line to determine the specific placement position of the chip pin; The initial position formed by the scanning line is set by calculating the center position of the characters on the chip surface to set an initial coordinate threshold, and further set an interval threshold between the formed scanning lines, and the specific effect of the threshold setting is: Firstly, the threshold setting can avoid unnecessary scanning line formation, thereby saving the consumption of computer memory and further saving time consumption; Secondly, the interval setting between the scanning lines can effectively avoid the wrong judgment caused by the different lengths of the chip pins, and also saves the scanning time of the computer and further improves the running speed of the algorithm; Thirdly, in the scanning process, the specific position of the chip pin can be determined by the threshold setting, and further the function of pin measurement can be realized, and according to the front and back comparison of the intersection result of the scanning line and the pin, whether the chip pin is missing can be determined, and the defect detection function of the chip pin is realized.

5. The method of claim 4, wherein: The defect detection function of the chip pin is to mark the corresponding result when the corresponding image preprocessing is performed, so as to realize the effect of defect classification: Firstly, when the best processing channel is selected, the distribution of the brightest pixels of the characters is calculated, and the calculated result can be used to determine whether the characters on the surface are printed full or light, if the characters are printed full, the value of the brightest pixels calculated is higher, otherwise, the result value obtained is lower; Secondly, when the connected domain of the character number is counted, whether the characters printed on the current chip exist the phenomenon of character loss can be determined according to the current statistical result; Thirdly, before the whole character connected domain is obtained, the whole character needs to be processed by difference, and the difference between the standard character and the character to be recognized is processed, and the result of the difference can determine whether the current character is few strokes, many strokes or broken; In the statistical classification of defect classification, all the algorithm marking is carried out in the necessary image preprocessing, and large-scale neural network training is not needed, so that a good and fast classification and recognition effect is achieved.

Images