CN112150541A - Multi-LED wafer positioning algorithm - Google Patents

Multi-LED wafer positioning algorithm Download PDF

Info

Publication number
CN112150541A
CN112150541A CN202010949813.0A CN202010949813A CN112150541A CN 112150541 A CN112150541 A CN 112150541A CN 202010949813 A CN202010949813 A CN 202010949813A CN 112150541 A CN112150541 A CN 112150541A
Authority
CN
China
Prior art keywords
image
led wafer
matched
searched
led
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010949813.0A
Other languages
Chinese (zh)
Inventor
宋华军
韩旭
李翼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China University of Petroleum East China
Original Assignee
China University of Petroleum East China
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China University of Petroleum East China filed Critical China University of Petroleum East China
Priority to CN202010949813.0A priority Critical patent/CN112150541A/en
Publication of CN112150541A publication Critical patent/CN112150541A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Container, Conveyance, Adherence, Positioning, Of Wafer (AREA)

Abstract

The invention discloses a multi-LED wafer positioning algorithm, which relates to the technical field of LED wafer positioning and comprises the following steps: the CCD camera shoots a local image on the LED wafer disc as an image to be positioned according to a certain step, and the LED wafer on the image to be positioned is positioned every time the local image is shot; when the LED wafer is positioned for each image to be positioned, matching positioning is carried out from the top of the image pyramid by adopting an edge gradient matching algorithm, and then mapping layer by layer and accurately positioning until the bottom layer of the pyramid is finally positioned; in the process of matching and positioning the pyramid top layer, a row of LED wafers or a column of LED wafers of the image to be positioned is accurately positioned, and then specific positioning of the remaining wafers in the image to be positioned currently is performed according to the measured row coordinates of each LED wafer in the row or column coordinates of each LED wafer in the column. Based on the prior art, the invention improves the speed and the accuracy of the LED wafer positioning algorithm and has better robustness.

Description

Multi-LED wafer positioning algorithm
Technical Field
The invention relates to the technical field of LED wafer positioning, in particular to a multi-LED wafer positioning algorithm.
Background
In the production process of the LED wafer, the LED wafer needs to be sorted, and the whole process comprises the steps of scanning and positioning the LED wafer and sorting and placing the LED wafer. For the positioning of the LED wafer, a CCD camera and software are usually used to achieve precise chip positioning using machine vision techniques. Semiconductor technologies 2013,38(7) based on pattern matching LED chip positioning system [ J ]; 556 and 559, in the above, precise positioning is realized by performing full-width fine matching only once, and positioning errors are caused by vibration generated by high-speed start and stop of the LED wafer sorting platform and interference of friction; in the Xia Junbo. template matching algorithm based on gradient search in the graph [ C ]// International reference on mechanics and Control (ICMC),2014: 1462-. But the processing time cannot meet the requirements of the visual servo system. Due to the particularity of the LED chip, the LED chip has a small size and a large number, and in the LED sorting platform, the accuracy and the matching speed need to be considered at the same time, so the existing methods are not suitable.
Disclosure of Invention
The invention aims to: a positioning technique suitable for multiple wafers and improving the matching speed is provided.
The technical scheme of the invention is as follows:
a multi-LED wafer positioning algorithm comprising the steps of:
the CCD camera shoots a local image on the LED wafer disc as an image to be positioned according to a certain step, and the LED wafer on the image to be positioned is positioned every time the local image is shot;
when the LED wafer is positioned for each image to be positioned, matching positioning is carried out from the top of the image pyramid by adopting an edge gradient matching algorithm, and then mapping layer by layer and accurately positioning until the bottom layer of the pyramid is finally positioned;
in the process of matching and positioning the pyramid top layer, a row of LED wafers or a column of LED wafers of the image to be positioned is accurately positioned, and then specific positioning of the remaining wafers in the image to be positioned currently is performed according to the measured row coordinates of each LED wafer in the row or column coordinates of each LED wafer in the column.
Further, in the image matching process, the selection steps of the image to be matched are as follows:
A. the camera is aligned to the center of the LED wafer disc, images to be matched are shot, and the images are detected;
B. after the detection is finished, the image to be matched is continuously moved to the right for a certain distance, and the detection is carried out once every time the image to be matched is moved until the image reaches the edge of the LED wafer disc;
C. moving the image to be matched up for a certain distance once, and detecting the image;
D. after the detection is finished, the image to be matched continuously moves a corresponding distance to the left, and the detection is carried out once every time the image to be matched moves until the edge of the LED wafer disc is reached;
E. repeating the step C and the step D until all images to be matched on the upper half part of the LED wafer disc are detected to be selected;
in the process, the movement of the image to be matched is realized according to the movement of a motor at the bottom of the LED wafer disc, and the motor moves to drive the LED wafer disc to move so as to move the position of the image to be matched, which is shot by the CCD camera;
the corresponding distance is set in advance according to the size of the image to be matched, and the detected position of the LED wafer has no repeated part after the image to be matched is shot each time;
after all the images to be matched on the upper half part of the LED wafer disc are selected, the images to be matched on the lower half part of the LED wafer disc are selected according to a similar method.
Specifically, the specific steps for detecting the upper half part of the LED wafer disc are as follows:
s1: obtaining LED wafer template image ItAnd an image I to be searched of the LED wafers；
S2: obtaining template images I with different rotation angles and pyramid layer numbertiAnd images I to be searched with different pyramid layer numberssi；
S3: method for solving LED wafers with different rotation angles and different pyramid levels by using Canny operatorTemplate image ItiThe edge points are used for solving the LED wafer template images I of different pyramid levels by the same methodsiThe edge points of (1);
s4: regarding edge points of the LED wafer template image as a point set pi＝(ri，ci)TEach point (r)i，ci) Corresponding gradient vector is di＝(ti，ui)TThe same processing is carried out on the image to be searched of the LED wafer to obtain each point (x, y) and the corresponding gradient vector e thereofx，y＝(vx，y，ωx，y)T(ii) a Affine transformation is carried out on the LED wafer template image to obtain a point set pi′＝ApiGradient vector di′＝(A-1)Tdi；
S5: acquiring the specific position of the LED wafer on the last line of the image to be searched; specifically, the gradient information similarity is calculated from the top layer of the image pyramid, after the approximate position similar to the template image is obtained at the top layer, the approximate position is mapped layer by layer and is accurately positioned until the bottom layer of the image pyramid is reached;
in the process of calculating the similarity at the top layer of the pyramid, traversing the image to be searched from left to right and from bottom to top, after calculating the matching degree each time, comparing the score with a preset matching degree threshold value T of the previous sub-image, if the score is higher than the threshold value T, temporarily storing the current sub-image position or replacing the coordinate and the rotating angle of the previous sub-image position, taking the threshold value as the threshold value of the next sub-image, if the matching degree of the next sub-image is lower than the threshold value of the previous sub-image, recording the coordinate and the rotating angle of the currently stored sub-image position, setting the edge gradient amplitude of the area where the matching target is located to be 0, and then matching the next position; recording the number of currently matched sub-images, and switching to S6 when the number of currently matched sub-images is equal to the number of the last line on the LED wafer image to be searched;
s6: matching from bottom to top according to the X coordinate and the rotation angle of each LED wafer on the last row;
s7: and outputting to obtain all wafer positions and rotation angles on the current image to be matched.
Specifically, the method for determining the number of the last line on the wafer image to be searched by the LED comprises the following steps: the LED wafer tray is determined in advance according to the fixed coordinates of each wafer when the LED wafer tray leaves the factory.
Specifically, the specific step of S1 is: graying the input original LED wafer template image and the image to be searched of the LED wafer, and performing Gaussian filtering by using a 3 x 3 Gaussian filter to obtain an LED wafer template image I after image preprocessingtAnd an image I to be searched of the LED wafers。
Specifically, the specific steps of S3 are as follows:
s31: calculating LED wafer template image ItiAnd an image I to be searched of the LED wafersiGradient intensity of each pixel
Figure BDA0002676552060000041
And direction of gradient
Figure BDA0002676552060000042
And the gradient direction of each pixel point is replaced by four directions of 0 degree, 45 degrees, 90 degrees and 135 degrees according to the similar degree;
s32: through LED wafer template image ItiAnd an image I to be searched of the LED wafersiSelecting edge points by the gradient strength of each pixel;
s33: the edges of the connections are tracked using dual threshold detection and hysteresis.
Specifically, in the step S32, the LED wafer template image ItiAnd an image I to be searched of the LED wafersiThe gradient strength of each pixel is compared with two pixels in the positive and negative gradient directions; if the gradient value of the current pixel is the largest compared with the gradient values of the other two pixels, the pixel point is reserved as an edge point, otherwise, the pixel point is not regarded as an edge point and is not reserved.
Specifically, the specific step of S33 includes setting a high threshold ThAnd a low threshold TlWill be less than the low threshold TlWill be greater than a high threshold TlPoint determination ofIf the edge point is a strong edge point, the point which is smaller than the high threshold and larger than the low threshold is a weak edge point; searching whether the 8 fields of the strong edge points have weak edge points, if so, continuing searching by taking the weak edge points as the center until the weak edge points cannot be searched; new edges are collected from this weak edge point again until the entire edge is closed.
Specifically, in S5, the similarity metric value is calculated by the following formula:
Figure BDA0002676552060000051
wherein, the S similarity metric value, n is the total number of template edge points, q represents the pixel point of the image to be searched, p represents the template image edge point, t'iRepresenting the x-direction gradient component of the edge points of the template image,
Figure BDA0002676552060000052
representing x-direction gradient component, u 'of edge point of sub-image to be searched'iRepresents the y-direction gradient component of the edge points of the template image,
Figure BDA0002676552060000053
representing the y-direction gradient component of the edge point of the sub-image to be searched;
setting a threshold S during the calculationminAs a stopping parameter; the dot product sum of the front j elements of the LED wafer template image is SjThe similarity measure of which is expressed as
Figure BDA0002676552060000054
When partial sum satisfies Sj＜S min1+ j/n, discarding the calculation of the similarity measure between the image to be searched and the template image after the jth element.
After the scheme is adopted, the invention has the following beneficial effects:
(1) the invention divides the wafer on the LED wafer into two parts, the positions of the first line or the last line of the LED wafer of the two parts are used as reference positions, and the positions of other wafers on the image to be matched are found according to the reference positions.
(2) In the invention, when traversing and matching a target, the edge gradient amplitude in the region where the matched target is located is set as 0, and the corresponding position of each pyramid level is processed, so that the score is low when the periphery is matched, the chip is not repeatedly identified near one chip, the sub-image region is not matched, the repeated matching is avoided, and the robustness of the algorithm is improved.
Drawings
FIG. 1 is a schematic diagram of an image to be searched of an LED wafer on an LED wafer tray according to an embodiment of the present invention;
FIG. 2 is a drawing of a template of an LED wafer in an embodiment of the present invention;
FIG. 3 is a diagram of an LED wafer to be searched without rotation in an embodiment of the present invention;
FIG. 4 is a diagram of the matching result without rotation in the embodiment of the present invention;
FIG. 5 is a diagram of an LED wafer to be searched with a 2.5 degree rotation in an embodiment of the present invention;
fig. 6 is a matching result diagram in the case of no rotation of 2.5 degrees in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention that are generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present invention, it should be noted that, if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings or the orientations or positional relationships that the products of the present invention are usually placed in when used, the terms are only used for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements indicated must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention.
Furthermore, the appearances of the terms "first," "second," "third," and the like, if any, are only used to distinguish one description from another, and are not to be construed as indicating or implying relative importance.
Furthermore, the terms "horizontal", "vertical", "overhang" and the like do not require that the components be absolutely horizontal or overhang, but may be slightly inclined. For example, "horizontal" merely means that the direction is more horizontal than "vertical" and does not mean that the structure must be perfectly horizontal, but may be slightly inclined.
The present embodiment discloses a multi-LED wafer positioning algorithm. The algorithm adopts a brand new mode for positioning the LED wafer, firstly positions the LED wafer in a row or a column on the image to be positioned of the LED, and then finds the positions of the other LED wafers on the image to be positioned according to the row or the column of the LED wafer. The method comprises the following steps:
the CCD camera shoots a local image on the LED wafer disc as an image to be positioned according to a certain step, and the LED wafer on the image to be positioned is positioned every time the local image is shot;
when the LED wafer is positioned for each image to be positioned, matching positioning is carried out from the top of the image pyramid by adopting an edge gradient matching algorithm, and then mapping layer by layer and accurately positioning until the bottom layer of the pyramid is finally positioned;
in the process of matching and positioning the pyramid top layer, a row of LED wafers or a column of LED wafers of the image to be positioned is accurately positioned, and then specific positioning of the remaining wafers in the image to be positioned currently is performed according to the measured row coordinates of each LED wafer in the row or column coordinates of each LED wafer in the column. In the actual operation process, the LED wafer is scanned transversely, so the present embodiment is described in a row-by-row manner, and if the LED wafer is scanned in a column-by-row manner, the principle is completely similar to the row-by-row manner, except that the moving direction of the motor is changed, and the moving direction of the sub-image selection during image matching is the same.
In the image matching process, the selection steps of the images to be matched are as follows:
A. the camera is aligned to the center of the LED wafer disc, images to be matched are shot, and the images are detected;
B. after the detection is finished, the image to be matched is continuously moved to the right for a certain distance, and the detection is carried out once every time the image to be matched is moved until the image reaches the edge of the LED wafer disc;
C. moving the image to be matched up for a certain distance once, and detecting the image;
D. after the detection is finished, the image to be matched continuously moves a corresponding distance to the left, and the detection is carried out once every time the image to be matched moves until the image reaches the edge of the LED wafer disc;
E. repeating the step C and the step D until all images to be matched on the upper half part of the LED wafer disc are detected to be selected;
in the process, the movement of the image to be matched is realized according to the movement of a motor at the bottom of the LED wafer disc, and the motor moves to drive the LED wafer disc to move so as to move the position of the image to be matched, which is shot by the CCD camera;
the corresponding distance is set in advance according to the size of the image to be matched, and the detected position of the LED wafer has no repeated part after the image to be matched is shot each time;
after all the images to be matched on the upper half part of the LED wafer disc are selected, the images to be matched on the lower half part of the LED wafer disc are selected according to a similar method. The selection of the image to be matched at the lower half part is as follows:
A1. the motor returns to the initial position, namely the CCD camera is aligned to the initial position of the LED wafer, the corresponding distance is left for the image to be matched, and the image is detected;
B1. after the detection is finished, the image to be matched continuously moves a corresponding distance to the left, and the detection is carried out once every time the image to be matched moves until the image reaches the edge of the LED wafer disc;
C1. moving the image to be matched down by a corresponding distance once, and detecting the image;
D1. after the detection is finished, the right phase of the image to be matched is continuously shifted by a corresponding distance, and the detection is carried out once every time the image to be matched is moved until the edge of the LED wafer disc is reached;
E1. and repeating the step C1 and the step D1 until all the lower half parts of the LED wafer disc to be matched are detected to be completely selected.
In the embodiment, there are 5 rows and 5 columns of wafers on each image to be matched, and when the next sub-image changes, the LED wafer tray relatively moves the positions of 5 LED wafers in the X direction or the Y direction, so that the wafers on the image to be matched are not missed or redundant.
It should be understood that the image to be matched does not move because the LED wafer tray is moving, the movement of the LED wafer tray is controlled by the motor, and the motor records the initial position of the first part, and after the detection of the upper part is completed, the motor can move to the initial position directly according to the record, and then the detection of the lower part is performed. The selection of the image to be matched is performed according to an S-shape no matter in the upper half part or the lower half part.
The core idea of the invention is that: for the images to be matched on the upper half parts of all the LED wafer disks, the number of the LED wafers in the last row is the largest, the LED wafers in the top row reach the edge part of the LED wafer disks, due to the distribution rule of the LED wafers, the LED wafers in the top row are lost, if the detection is started from the top, the detection is missed, but if the wafers in the last row are detected, the situation is not caused. Similarly, when the wafer tray at the lower half part is detected, the number of the LED wafers in the first row of each image to be matched is the largest, so that the LED wafer tray is roughly divided into an upper part and a lower part. The number of the LED wafers on the actual LED wafer disc is large, and most of the LED wafers in the middle of the image to be detected of the LED wafer are not lost, and only the edge parts are different.
The following will explain in detail the LED wafer positioning algorithm of the upper half of the LED wafer tray, which includes the following specific steps:
s1: graying the input original LED wafer template image (shown in figure 2) and the LED wafer image to be searched (shown in figure 3), and performing Gaussian filtering by using a 3 x 3 Gaussian filter to obtain an LED wafer template image I after image preprocessingtAnd an image I to be searched of the LED wafers。
S2: to LED wafer template image ItPerforming a series of rotations at a certain step length, and performing pyramid downsampling on the LED wafer template image at each rotation angle to generate a series of template images I with different rotation angles and pyramid layer numberti(ii) a To-be-searched image I of LED wafersPyramid downsampling is carried out to generate images I to be searched with different pyramid layer numberssi。
S3: obtaining LED wafer template images I with different rotation angles and different pyramid levels by using Canny operatortiThe edge points are used for solving the LED wafer template images I of different pyramid levels by the same methodsiThe edge points of (1); the specific steps of S3 are as follows:
s31: calculating LED wafer template image ItiAnd an image I to be searched of the LED wafersiGradient intensity of each pixel
Figure BDA0002676552060000111
And direction of gradient
Figure BDA0002676552060000112
And the gradient direction of each pixel point is replaced by four directions of 0 degree, 45 degrees, 90 degrees and 135 degrees according to the similar degree;
s32: through LED wafer template image ItiAnd an image I to be searched of the LED wafersiSelecting edge points by the gradient strength of each pixel; template image I of LED wafertiAnd an image I to be searched of the LED wafersiThe gradient strength of each pixel is compared with two pixels in the positive and negative gradient directions; if the gradient value of the current pixel is the largest compared with the other two pixel values, the pixel point is reserved as an edge point, otherwise, the pixel point is not regarded as an edge point and is not reserved.
S33: the edges of the connections are tracked using dual threshold detection and hysteresis. Setting a high threshold ThAnd a low threshold TlWill be less than the low threshold TlWill be greater than a high threshold TlDetermining the point of (2) as a strong edge point, and taking the point smaller than the high threshold value and larger than the low threshold value as a weak edge point; searching whether the 8 fields of the strong edge points have weak edge points, if so, continuing searching by taking the weak edge points as the center until the weak edge points cannot be searched; new edges are collected from this weak edge point again until the entire edge is closed.
S4: regarding edge points of the LED wafer template image as a point set pi＝(ri，ci)TEach point (r)i，ci) Corresponding gradient vector is di＝(ti，ui)TThe same processing is carried out on the image to be searched of the LED wafer to obtain each point (x, y) and the corresponding gradient vector e thereofx，y＝(vx，y，ωx，y)T(ii) a Affine transformation is carried out on the LED wafer template image to obtain a point set pi′＝ApiGradient vector di′＝(A-1)Tdi；
S5: acquiring the specific position of the LED wafer in the last row (if the lower half part is true, the first row) of the image to be searched; specifically, the gradient information similarity is calculated from the top layer of the image pyramid, after the approximate position similar to the template image is obtained at the top layer, the approximate position is mapped layer by layer and is accurately positioned until the bottom layer of the image pyramid is reached; in S5, the similarity metric value is calculated by the following formula:
Figure BDA0002676552060000121
wherein, the S similarity metric value, n is the total number of template edge points, q represents the pixel point of the image to be searched, p represents the template image edge point, t'iRepresenting the x-direction gradient component of the edge points of the template image,
Figure BDA0002676552060000122
representing x-direction gradient component, u 'of edge point of sub-image to be searched'iRepresents the y-direction gradient component of the edge points of the template image,
Figure BDA0002676552060000123
representing the y-direction gradient component of the edge point of the sub-image to be searched;
setting a threshold S during the calculationminAs a stopping parameter; the dot product sum of the front j elements of the LED wafer template image is SjThe similarity measure of which is expressed as
Figure BDA0002676552060000131
When partial sum satisfies Sj＜S min1+ j/n, discarding the calculation of the similarity measure between the image to be searched and the template image after the jth element.
In the process of calculating the similarity at the pyramid top layer, traversing the image to be searched from left to right and from bottom to top (if the similarity is lower half, the similarity is from left to right and from top to bottom), after calculating the matching degree each time, comparing the score with a preset matching degree threshold value T of a previous sub-image, if the score is higher than the threshold value T, temporarily storing the coordinate and the rotation angle of the current sub-image position or replacing the previous sub-image position, taking the threshold value as the threshold value of a next sub-image, if the matching degree of the next sub-image is lower than the threshold value of the previous sub-image, recording the coordinate and the rotation angle of the currently stored sub-image position, setting the edge gradient amplitude of the area where the matching target is located to be 0, and then matching the next position; recording the number of currently matched subgraphs, and switching to S6 when the number of currently matched subgraphs is equal to the number of the last line (if the lower half part is the first line) on the LED wafer image to be searched;
s6: matching is performed from bottom to top (from top to bottom if the lower half) according to the X coordinate and the rotation angle of each LED chip on the last row (the first row if the lower half), and the matching process is similar to S5; the method comprises the following specific steps: calculating gradient information similarity from the top layer of the image pyramid, obtaining approximate positions similar to the template image at the top layer, mapping layer by layer, and accurately positioning until the bottom layer of the image pyramid; in the process of calculating the similarity at the top layer of the pyramid, selecting sub-images with the same size as the image template, traversing the image to be searched from the specific X coordinate of the LED wafer in the first row according to a certain step length, calculating the matching degree each time, comparing the score with a preset matching degree threshold value T of the previous sub-image, if the score is higher than the threshold value T, temporarily storing the current sub-image position or replacing the coordinate and the rotating angle of the previous sub-image position, taking the threshold value as the threshold value of the next sub-image, if the matching degree of the next sub-image is lower than the threshold value of the previous sub-image, recording the coordinate and the rotating angle of the currently stored sub-image position, and then matching the next position; when the LED wafer positioning in one row is completed, the next row of LED wafer positioning is performed.
S7: and outputting to obtain all wafer positions and rotation angles on the current image to be matched.
The method for determining the number of the last line on the LED wafer image to be searched comprises the following steps: the LED wafer tray is determined in advance according to fixed coordinates of each wafer of the LED wafer tray which leaves a factory and the moving position of the motor. The specific principle is as follows:
when the LED wafer leaves a factory, the factory gives the positions of all the LED wafers, selects a central chip, takes the position of the chip as the central position, and gives the coordinates of (0,0) and the positions of all the other chips, and the coordinates are given in the form of a table file. The chip coordinates of the positions adjacent to the left and right of the center chip are (1,0) (-1,0), and it is noted that these position coordinates are not the position coordinates to be detected by the present invention. In the LED wafer positioning process of the LED wafer disc, the full disc chip positioning is started by taking the (0,0) chip as the first image center chip. In this case, as shown in fig. 1, the content enclosed by the left frame is 9 chips enclosed in fig. 1, and the chips with coordinates (0,0) are marked in the figure (note that the coordinates refer to the coordinates of the (0,0) chip in the table, that is, the relative coordinates). Recording the Y coordinate of the center chip as 0, and finding out the maximum value 1 (representing 3 in the image) and the minimum value-1 when Y is 1 (representing the first line of the image) and Y is-1 (representing the last line of the image) from the table; a total of 3 values for-1, 0, 1, indicating that there are at most 3 wafers in a row at this time. After the image to be processed is processed, the image is shifted to the right by a distance of 3 chips (actually, the LED wafer is driven by the motor to move to the left), as shown in the content outlined by the right frame in fig. 1, at this time, the boundary of the LED wafer is reached, and the maximum X value is 3 and the minimum X value is 2 when Y is 0-1 or-1, which indicates that there are 2 LED wafers in the last row. And so on. Similarly, the lower semicircle can be used in the above way, but only the first row needs to be seen.
Obviously, for the purpose of illustration, the number of wafers on the LED wafer tray, the number of LED chips on the image to be detected, and the number of movements are all shown for convenience of illustration, and the actual size, number, and difference in the figure are that there are 25 chips to be detected in the image to be detected in the center, as shown in fig. 3.
After the scheme of the invention is adopted, the LED wafer template picture in figure 2 is selected, wherein the image to be detected at one time is shown in figure 3, and the matching result is shown in figure 4; for the image rotated by 2.5 degrees, the image to be detected at one time is shown in fig. 5, and the matching result is shown in fig. 6. As can be seen from fig. 4 and 6, the method of the present invention has high accuracy of the matching result. Further, the present invention and the HexSight software respectively perform the inspection on the same LED wafer disc, and some results are shown in table 1 and table 2. It can be seen that the recognition rate of the present invention is very high, and the deviation of the X-coordinate and the Y-coordinate does not exceed 1 pixel.
TABLE 1 results of different angle wafer recognition rates
Figure BDA0002676552060000151
TABLE 2 identification of coordinate comparison results (unit: pixel)
Figure BDA0002676552060000152
Figure BDA0002676552060000161
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. A multi-LED wafer positioning algorithm, comprising the steps of:
the CCD camera shoots a local image on the LED wafer disc as an image to be positioned according to a certain step, and the LED wafer on the image to be positioned is positioned every time the local image is shot;
when the LED wafer is positioned for each image to be positioned, matching positioning is carried out from the top of the image pyramid by adopting an edge gradient matching algorithm, and then mapping layer by layer and accurately positioning until the bottom layer of the pyramid is finally positioned;
in the process of matching and positioning the pyramid top layer, a row of LED wafers or a column of LED wafers of the image to be positioned is accurately positioned, and then specific positioning of the remaining wafers in the image to be positioned currently is performed according to the measured row coordinates of each LED wafer in the row or column coordinates of each LED wafer in the column.
2. The multi-LED wafer positioning algorithm of claim 1, wherein in the image matching process, the selection of the image to be matched is as follows:
A. the camera is aligned to the center of the LED wafer disc, images to be matched are shot, and the images are detected;
B. after the detection is finished, the image to be matched is continuously moved to the right for a certain distance, and the detection is carried out once every time the image to be matched is moved until the image reaches the edge of the LED wafer disc;
C. moving the image to be matched up for a certain distance once, and detecting the image;
D. after the detection is finished, the image to be matched continuously moves a corresponding distance to the left, and the detection is carried out once every time the image to be matched moves until the edge of the LED wafer disc is reached;
E. repeating the step C and the step D until all images to be matched on the upper half part of the LED wafer disc are detected to be selected;
in the process, the movement of the image to be matched is realized according to the movement of a motor at the bottom of the LED wafer disc, and the motor moves to drive the LED wafer disc to move so as to move the position of the image to be matched, which is shot by the CCD camera;
the corresponding distance is set in advance according to the size of the image to be matched, and the detected position of the LED wafer has no repeated part after the image to be matched is shot each time;
after all the images to be matched on the upper half part of the LED wafer disc are selected, the images to be matched on the lower half part of the LED wafer disc are selected according to a similar method.
3. The multi-LED wafer positioning algorithm of claim 1, wherein the specific steps of detecting the top half of the LED wafer tray are as follows:
s1: obtaining LED wafer template image ItAnd an image I to be searched of the LED wafers；
S2: obtaining template images I with different rotation angles and pyramid layer numbertiAnd images I to be searched with different pyramid layer numberssi；
S3: obtaining LED wafer template images I with different rotation angles and different pyramid levels by using Canny operatortiThe edge points are used for solving the LED wafer template images I of different pyramid levels by the same methodsiThe edge points of (1);
s4: regarding edge points of the LED wafer template image as a point set pi＝(ri，ci)TEach point (r)i，ci) Corresponding gradient vector is di＝(ti，ui)TThe same processing is carried out on the image to be searched of the LED wafer to obtain each point (x, y) and the corresponding gradient vector e thereofx，y＝(vx，y，ωx，y)T(ii) a Affine transformation is carried out on the LED wafer template image to obtain a point set pi′＝ApiGradient vector di′＝(A-1)Tdi；
S5: acquiring the specific position of the LED wafer on the last line of the image to be searched; specifically, the gradient information similarity is calculated from the top layer of the image pyramid, after the approximate position similar to the template image is obtained at the top layer, the approximate position is mapped layer by layer and is accurately positioned until the bottom layer of the image pyramid is reached;
in the process of calculating the similarity at the top layer of the pyramid, traversing the image to be searched from left to right and from bottom to top, after calculating the matching degree each time, comparing the score with a preset matching degree threshold value T of the previous sub-image, if the score is higher than the threshold value T, temporarily storing the current sub-image position or replacing the coordinate and the rotating angle of the previous sub-image position, taking the threshold value as the threshold value of the next sub-image, if the matching degree of the next sub-image is lower than the threshold value of the previous sub-image, recording the coordinate and the rotating angle of the currently stored sub-image position, setting the edge gradient amplitude of the area where the matching target is located to be 0, and then matching the next position; recording the number of currently matched sub-images, and switching to S6 when the number of currently matched sub-images is equal to the number of the last line on the LED wafer image to be searched;
s6: matching is carried out from bottom to top according to the X coordinate and the rotation angle of each LED wafer on the last row, and the matching process is similar to S5;
s7: and outputting to obtain all wafer positions and rotation angles on the current image to be matched.
4. The multi-LED wafer positioning algorithm of claim 3, wherein the method for determining the number of the last row on the wafer image to be searched by the LED comprises: the LED wafer tray is determined in advance according to the fixed coordinates of each wafer when the LED wafer tray leaves the factory.
5. The multi-LED wafer positioning algorithm according to claim 3, wherein the specific steps of S1 are: graying the input original LED wafer template image and the image to be searched of the LED wafer, and performing Gaussian filtering by using a 3 x 3 Gaussian filter to obtain an LED wafer template image I after image preprocessingtAnd an image I to be searched of the LED wafers。
6. The multi-LED wafer positioning algorithm according to claim 3, wherein the specific steps of S3 are as follows:
s31: calculating LED wafer template image ItiAnd an image I to be searched of the LED wafersiGradient intensity of each pixel
Figure FDA0002676552050000041
And direction of gradient
Figure FDA0002676552050000042
And the gradient direction of each pixel point is replaced by four directions of 0 degree, 45 degrees, 90 degrees and 135 degrees according to the similar degree;
s32: through LED wafer template image ItiAnd an image I to be searched of the LED wafersiSelecting edge points by the gradient strength of each pixel;
s33: the edges of the connections are tracked using dual threshold detection and hysteresis.
7. According toThe multi-LED wafer positioning algorithm of claim 6, wherein in S32, the LED wafer template image I istiAnd an image I to be searched of the LED wafersiThe gradient strength of each pixel is compared with two pixels in the positive and negative gradient directions; if the gradient value of the current pixel is the largest compared with the gradient values of the other two pixels, the pixel point is reserved as an edge point, otherwise, the pixel point is not regarded as an edge point and is not reserved.
8. The multi-LED wafer positioning algorithm of claim 6, wherein the specific step of S33 includes setting a high threshold ThAnd a low threshold TlWill be less than the low threshold TlWill be greater than a high threshold TlDetermining the point of (2) as a strong edge point, and taking the point smaller than the high threshold value and larger than the low threshold value as a weak edge point; searching whether the 8 fields of the strong edge points have weak edge points, if so, continuing searching by taking the weak edge points as the center until the weak edge points cannot be searched; new edges are collected from this weak edge point again until the entire edge is closed.
9. The multi-LED wafer positioning algorithm of claim 3, wherein in S5, the similarity metric value is calculated by:
Figure FDA0002676552050000043
wherein, the S similarity metric value, n is the total number of template edge points, q represents the pixel point of the image to be searched, p represents the template image edge point, t'iRepresenting the x-direction gradient component of the edge points of the template image,
Figure FDA0002676552050000051
representing x-direction gradient component, u 'of edge point of sub-image to be searched'iRepresents the y-direction gradient component of the edge points of the template image,
Figure FDA0002676552050000052
representing the y-direction gradient component of the edge point of the sub-image to be searched;
setting a threshold S during the calculationminAs a stopping parameter; the dot product sum of the front j elements of the LED wafer template image is SjThe similarity measure of which is expressed as
Figure FDA0002676552050000053
When partial sum satisfies Sj＜Smin1+ j/n, discarding the calculation of the similarity measure between the image to be searched and the template image after the jth element.
CN202010949813.0A 2020-09-10 2020-09-10 Multi-LED wafer positioning algorithm Pending CN112150541A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010949813.0A CN112150541A (en) 2020-09-10 2020-09-10 Multi-LED wafer positioning algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010949813.0A CN112150541A (en) 2020-09-10 2020-09-10 Multi-LED wafer positioning algorithm

Publications (1)

Publication Number Publication Date
CN112150541A true CN112150541A (en) 2020-12-29

Family

ID=73889664

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010949813.0A Pending CN112150541A (en) 2020-09-10 2020-09-10 Multi-LED wafer positioning algorithm

Country Status (1)

Country Link
CN (1) CN112150541A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114187294A (en) * 2022-02-16 2022-03-15 常州铭赛机器人科技股份有限公司 Regular wafer positioning method based on prior information
CN116863176A (en) * 2023-09-04 2023-10-10 苏州聚视兴华智能装备有限公司 Image template matching method for digital intelligent manufacturing

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000323395A (en) * 1999-05-13 2000-11-24 Sony Corp Aligner and method of flattening imaging-surface
US20040189674A1 (en) * 2003-03-31 2004-09-30 Zhengyou Zhang System and method for whiteboard scanning to obtain a high resolution image
CN101793843A (en) * 2010-03-12 2010-08-04 华东理工大学 Connection table based automatic optical detection algorithm of printed circuit board
CN104700085A (en) * 2015-03-10 2015-06-10 华中科技大学 Template matching-based chip positioning method
CN106127258A (en) * 2016-07-01 2016-11-16 华中科技大学 A kind of target matching method
CN106384113A (en) * 2016-11-07 2017-02-08 湖南源信光电科技有限公司 Odd-and-even-line license plate character dividing method based on projection and template matching
CN107480727A (en) * 2017-08-28 2017-12-15 荆门程远电子科技有限公司 The unmanned plane image fast matching method that a kind of SIFT and ORB are combined
CN109405755A (en) * 2018-12-13 2019-03-01 西安交通大学 A kind of large scale tube sheet aperture and pitch measurement device and measurement method
CN109668904A (en) * 2019-01-25 2019-04-23 南京理工大学 A kind of optical element flaw inspection device and method
CN109871027A (en) * 2017-12-05 2019-06-11 深圳市九天创新科技有限责任公司 A kind of oblique photograph method and system
CN110378879A (en) * 2019-06-26 2019-10-25 杭州电子科技大学 A kind of Bridge Crack detection method
CN111640154A (en) * 2020-05-24 2020-09-08 西安交通大学 Vertical needle micro-plane sub-pixel level positioning method based on micro-vision

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000323395A (en) * 1999-05-13 2000-11-24 Sony Corp Aligner and method of flattening imaging-surface
US20040189674A1 (en) * 2003-03-31 2004-09-30 Zhengyou Zhang System and method for whiteboard scanning to obtain a high resolution image
CN101793843A (en) * 2010-03-12 2010-08-04 华东理工大学 Connection table based automatic optical detection algorithm of printed circuit board
CN104700085A (en) * 2015-03-10 2015-06-10 华中科技大学 Template matching-based chip positioning method
CN106127258A (en) * 2016-07-01 2016-11-16 华中科技大学 A kind of target matching method
CN106384113A (en) * 2016-11-07 2017-02-08 湖南源信光电科技有限公司 Odd-and-even-line license plate character dividing method based on projection and template matching
CN107480727A (en) * 2017-08-28 2017-12-15 荆门程远电子科技有限公司 The unmanned plane image fast matching method that a kind of SIFT and ORB are combined
CN109871027A (en) * 2017-12-05 2019-06-11 深圳市九天创新科技有限责任公司 A kind of oblique photograph method and system
CN109405755A (en) * 2018-12-13 2019-03-01 西安交通大学 A kind of large scale tube sheet aperture and pitch measurement device and measurement method
CN109668904A (en) * 2019-01-25 2019-04-23 南京理工大学 A kind of optical element flaw inspection device and method
CN110378879A (en) * 2019-06-26 2019-10-25 杭州电子科技大学 A kind of Bridge Crack detection method
CN111640154A (en) * 2020-05-24 2020-09-08 西安交通大学 Vertical needle micro-plane sub-pixel level positioning method based on micro-vision

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
XIA JUN-BO: "Template matching algorithm based on gradient search", 《2014 INTERNATIONAL CONFERENCE ON MECHATRONICS AND CONTROL (ICMC)》 *
刘诗浩: "基于视觉的LED晶片位置识别算法研究", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 *
李德龙等: "基于视觉伺服的LED芯片形状匹配算法研究", 《计算机工程与应用》 *
杨诚: "LED晶圆片的快速模板匹配算法研究", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114187294A (en) * 2022-02-16 2022-03-15 常州铭赛机器人科技股份有限公司 Regular wafer positioning method based on prior information
CN116863176A (en) * 2023-09-04 2023-10-10 苏州聚视兴华智能装备有限公司 Image template matching method for digital intelligent manufacturing
CN116863176B (en) * 2023-09-04 2023-12-05 苏州聚视兴华智能装备有限公司 Image template matching method for digital intelligent manufacturing

Similar Documents

Publication Publication Date Title
US11227405B2 (en) Determining positions and orientations of objects
CN110315525A (en) A kind of robot workpiece grabbing method of view-based access control model guidance
EP1678659B1 (en) Method and image processing device for analyzing an object contour image, method and image processing device for detecting an object, industrial vision apparatus, smart camera, image display, security system, and computer program product
JP5339065B2 (en) Object tracking device
CN111126174A (en) Visual detection method for robot to grab parts
CN112150541A (en) Multi-LED wafer positioning algorithm
US6690819B1 (en) Method and apparatus for recognizing components
CN115816471B (en) Unordered grabbing method, unordered grabbing equipment and unordered grabbing medium for multi-view 3D vision guided robot
CN106067031A (en) Cooperate with the degree of depth learning network Machine Vision Recognition system based on artificial mechanism for correcting errors
CN109711457A (en) It is a kind of based on improve the HU not rapid image matching method of bending moment and its application
Czyzewski et al. Chessboard and chess piece recognition with the support of neural networks
CN112329880A (en) Template fast matching method based on similarity measurement and geometric features
US7055112B2 (en) Method for automatically defining a part model for semiconductor components
CN112338898B (en) Image processing method and device of object sorting system and object sorting system
CN116309573B (en) Defect detection method for printed characters of milk packaging box
CN113505629A (en) Intelligent storage article recognition device based on light weight network
CN111199533B (en) Image processing apparatus and method
CN109741302B (en) SD card form recognition system and method based on machine vision
JP5160366B2 (en) Pattern matching method for electronic parts
JP6041710B2 (en) Image recognition method
CN116228783A (en) Workpiece point cloud segmentation method under complex background and application thereof
JP3993044B2 (en) Appearance inspection method, appearance inspection device
JP4566686B2 (en) Method and apparatus for determining shape of object
JPH0953915A (en) Method for recognizing overlapping state
CN110544240A (en) patch position error detection method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination