CN108596875B - Image segmentation algorithm-based semiconductor chip flash rapid detection method - Google Patents

Image segmentation algorithm-based semiconductor chip flash rapid detection method Download PDF

Info

Publication number
CN108596875B
CN108596875B CN201810235175.9A CN201810235175A CN108596875B CN 108596875 B CN108596875 B CN 108596875B CN 201810235175 A CN201810235175 A CN 201810235175A CN 108596875 B CN108596875 B CN 108596875B
Authority
CN
China
Prior art keywords
similarity
region
image
size
gaussian
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.)
Active
Application number
CN201810235175.9A
Other languages
Chinese (zh)
Other versions
CN108596875A (en
Inventor
赵春晖
王晶
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.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
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 Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201810235175.9A priority Critical patent/CN108596875B/en
Publication of CN108596875A publication Critical patent/CN108596875A/en
Application granted granted Critical
Publication of CN108596875B publication Critical patent/CN108596875B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • 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
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G06T7/44—Analysis of texture based on statistical description of texture using image operators, e.g. filters, edge density metrics or local histograms
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30148—Semiconductor; IC; Wafer

Abstract

The invention discloses a method for quickly detecting flash of a semiconductor chip based on an image segmentation algorithm, which aims to provide timely feedback information for an EMC (electro magnetic compatibility) packaging process by identifying surplus substances overflowing to pins and exposed carriers in the EMC packaging process. Firstly, preprocessing an original image of an EMC packaged and formed acquired by an industrial camera, then carrying out image segmentation by using a Selective search algorithm to obtain a target candidate region, screening a chip region to be detected from the target candidate region, and carrying out threshold segmentation on the screened region to obtain a flash region. The method can give the flash detection result of each chip in a very short time, not only optimizes the flash detection means in the EMC packaging process, but also provides real-time feedback information for the EMC packaging process, thereby improving the performance of the EMC packaging.

Description

Image segmentation algorithm-based semiconductor chip flash rapid detection method
Technical Field
The invention belongs to the field of integrated circuit EMC packaging flash detection, and particularly relates to a semiconductor chip flash rapid detection method based on an image segmentation algorithm.
Background
EMC is an important microelectronic packaging material, and a semiconductor chip is coated and protected by a packaging process so as to be prevented from being damaged by the external environment, and meanwhile, the EMC also has a certain heat dissipation effect. The EMC has the characteristics of large-scale production and reasonable reliability, is one of common packaging materials for semiconductor packaging, and occupies more than 95% of packaging markets by virtue of unique advantages. However, in the semiconductor plastic package process, because a gap exists between the lead frame and the plastic package mold, the colloidal plastic package material leaks out from the gap and is solidified on the surface of the lead frame to form a black flash, which is an excess on the integrated circuit pins and the exposed carrier. The flash does not have adverse effect on the performance of the plastic package product, but because the flash covers the terminal pins, if the flash is not processed, an insulation area is formed to affect the subsequent electroplating quality and the product reliability, and the problems of product open circuit, cold welding and the like are caused. Therefore, how to reduce the occurrence of flash and the method of removing flash have been an important problem for researchers and manufacturers.
The reason why EMC package flash is generated is generally: 1) the plastic package material has good fluidity and low viscosity, and the plastic package material/epoxy resin can overflow from a parting surface to cover the terminal pin under the condition of loose die assembly; 2) after the plastic package mold is used for a long time, the surface of the plastic package mold is abraded or the base is not flat, so that flash is caused; 3) the frame precision is not matched with the mould in the semiconductor packaging process, and the edge pressing problem is caused by insufficient meshing. After the flash is formed, the flash itself has no influence on the performance of the plastic package product, but since the flash covers the terminal pins, plating defects may be formed to affect the reliability of the product. At present, manufacturers generally carry out remediation by a method of removing residual flash through a file, sanding and weak acid cleaning after a flash removing process, but the method is easy to scratch a plastic package body and cause the surface of a terminal pin to be rough, so that the appearance quality and the electroplating quality are influenced, and the production progress is also influenced. Therefore, it is important to avoid and reduce the generation of flash. In conclusion, a visual green pollution-free flash detection method is developed, so that the flash condition of a chip can be quickly detected, the requirement of providing reference basis for the EMC packaging process in time is met, and the method has important practical significance.
Thanks to the development of computer technology and image processing technology, a novel visual flash detection technology appears, the detection result of which is a gray image, and the flash detection technology has the characteristics of high detection precision, flexible use, large information amount and the like, and is very suitable for flash detection in the chip packaging production process.
Disclosure of Invention
The invention aims to provide a rapid flash detection strategy aiming at the flash problem in the existing EMC packaging process. The strategy can accurately evaluate the chip flash situation on site, so as to provide feedback information for adjustment and maintenance of in-service equipment, and is an important reference basis for an EMC packaging process.
The purpose of the invention is realized by the following technical scheme: a semiconductor chip flash rapid detection method based on an image segmentation algorithm comprises the following steps:
a semiconductor chip flash rapid detection method based on an image segmentation algorithm is characterized by comprising the following steps:
(1) producing a packaging original sheet by using a semiconductor packaging press (FSAM120-1US) and a QFNQFNB 7X7-48L (T0.75) mould, and acquiring a multi-frame gray image of the packaging original sheet by using an industrial camera;
(2) performing Gaussian blur preprocessing on each frame of image;
(3) based on selective search image segmentation, hundreds of candidate regions are obtained and chip regions to be analyzed are screened out, and the method is realized by the following substeps:
(3.1) over-dividing the single-frame image to obtain n initial regions r1,r2,...,rnForming an initial region set R ═ { R ═ R1,r2,...,rn}; meanwhile, an initialization similarity set S is constructed,
Figure BDA0001603773750000021
(3.2) calculating the similarity s (r) of every two adjacent areasi,rj) The similarity index adopts color similarity, size similarity, texture similarity, coincidence similarity and four complementary similarities, and the four complementary similarities are respectively combined for the areas from different aspects;
(a) the color similarity is calculated as follows:
dividing each initial region into q intervals according to pixel values to obtain a normalized color histogram
Figure BDA0001603773750000022
Wherein the content of the first and second substances,
Figure BDA0001603773750000023
a normalization frequency count representing a kth pixel value interval in the ith initial region;
two adjacent regions ri,rjThe color similarity of (a) is:
Figure BDA0001603773750000024
merge ri,rjThe new region r obtainedijThe color histogram is:
Figure BDA0001603773750000025
size(rij)＝size(ri)+size(rj) (19)
wherein size (r)i) Is an initial region riThe size of (d);
(b) the texture similarity calculation method comprises the following steps:
calculating the SIFT (scale-invariant feature transform) feature of each initial region, wherein the value of a scale factor in the patent is 1, and the width of a histogram interval is 10 to obtain the SIFT feature
Figure BDA0001603773750000031
Obtained as a textural feature.
Wherein the content of the first and second substances,
Figure BDA0001603773750000032
representing the kth texture feature in the ith initial region;
two adjacent regions ri,rjThe texture similarity of (a) is:
Figure BDA0001603773750000033
merge ri,rjThe new region r obtainedijThe histogram of the texture is
Figure BDA0001603773750000034
(C) The calculation method of the size similarity comprises the following steps:
Figure BDA0001603773750000035
where size (im) represents the size of the entire grayscale image.
(d) The calculation method of the coincidence similarity comprises the following steps:
Figure BDA0001603773750000036
wherein BBijTo surround only the region riAnd rjThe bounding box of (1).
And finally, integrating the four similarities to obtain the final similarity:
s(ri,rj)＝a1Scolor(ri,rj)+a2Stexture(ri,rj)+a3Ssize(ri,rj)+a4Sfill(ri,rj)(24)
wherein a isi∈ {0,1}, indicating whether or not the corresponding similarity is used.
(3.3) all the similarities form a similarity set S, and the two adjacent regions S (r) with the highest similarity in the similarity seti,rj) Max(s), merge the two regions rij＝ri∪rjCalculating the new region rijAnd (3) updating the similarity set with the similarity of adjacent regions: deleting the region r with which the data is mergediAll similarities s (r) of the correlationi,r*) And with the region r being mergedjAll similarities s (r) of the correlationj,r#) And adding the updated similarity set as follows: adding a new region rijSimilarity to adjacent regions;
and (3.4) circulating the step 3.2-3.3, and ending the circulation when the similarity set S is an empty set. Completing the combination of the initial regions to obtain a candidate region set R;
(3.5) screening an upper chip area and a lower chip area to be analyzed from the candidate area set R, so that the size of the area to be analyzed is in the range of one third to one half of the size of the original packaging piece, and the two areas to be analyzed need to be respectively contained
Figure BDA0001603773750000041
And
Figure BDA0001603773750000042
the resolution of the 4 key geometric points and the packaging original sheet is a x b.
(4) Performing threshold-based image segmentation on the screened region to be analyzed, wherein the step is realized by the following substeps:
(4.1) calculating a normalized histogram of the region to be analyzed. Let {0, 1.,. L-1} denote L different gray levels in a region image of size M N pixels, NzIndicating the number of pixels with a gray level z. The total number of pixels in the area image is MN ═ n0+n1+n2+...+nL-1，pz＝nzL-1 represents a normalized frequency of z for the gray level in the histogram.
(4.2) for u-0, 1,21(u), the calculation formula is as follows:
Figure BDA0001603773750000043
(4.3) for u 0,1, 2.., L-1, calculate the cumulative mean m (u) as follows:
Figure BDA0001603773750000044
(4.4) calculating the Global Gray mean mGThe calculation formula is as follows:
Figure BDA0001603773750000045
(4.5) for u-0, 1,2
Figure BDA0001603773750000046
The calculation formula is as follows:
Figure BDA0001603773750000047
(4.6) obtaining the Otsu threshold u by maximizing the objective function (13)*If the maximum values are not unique, u is obtained by averaging the respective detected maximum values u*；
Figure BDA0001603773750000051
(4.7) based on the optimal threshold u*The region image is segmented by the following formula to obtain a segmented image:
Figure BDA0001603773750000052
f (x, y) is the pixel of the pixel point (x, y) in the image of the area to be analyzed. And the segmented pixels are 1 and represent the detected flash part in the production process.
The invention has the beneficial effects that: the invention relates to a rapid flash detection method oriented to an EMC packaging process, which divides an image into hundreds to thousands of candidate areas through a selective search algorithm and screens out a chip area to be analyzed from the candidate areas. And further, the optimal threshold value is utilized to segment the image to be analyzed to obtain a flash image, so that whether the flash condition is serious or not can be judged. According to each frame of image, the method can provide a detection result within 20 seconds, can quickly feed back flash information in an EMC packaging process, and provides a reference basis, so that more flash conditions are avoided.
Drawings
FIG. 1(a) is a flow chart of the semiconductor package of the present invention, FIG. 1(b) is the parameter setting in the actual production operation, and FIG. 1(c) is the final package raw sheet;
FIG. 2 is a schematic diagram of the semiconductor package flash detection system of the present invention, wherein an industrial camera obtains an original gray image and transmits the image to a computer for further analysis;
FIG. 3 is a gray scale image of the original packaging sheet obtained by the present invention;
FIG. 4 is a reduced resolution image obtained by Gaussian blur preprocessing according to the method of the present invention, which retains image integrity information while reducing image resolution;
FIG. 5 is a partial result display obtained after the image is segmented by the selective search of the method of the present invention;
FIG. 6 shows the region to be analyzed selected by the method of the present invention;
FIG. 7 is a final flash image after thresholding by the method of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific examples.
(1) Emc (epoxy molding compound) package: as shown in fig. 1(a), the lead frame is placed in a mold so that each chip is located in an acupuncture point, and the block-shaped EMC is placed in the mold air after the mold is closed. Setting parameters of a packaging press according to the figure 1(b), melting EMC at high temperature, flowing to acupuncture points along a track, gradually covering the chip from the bottom until the chip is completely covered and wrapped, and molding and curing, as shown in figure 1 (c);
(2) image acquisition and preprocessing: the image-based flash detection system is shown in fig. 2, and an industrial camera is used for obtaining a gray scale image of a packaging original sheet shown in fig. 3, wherein the resolution is 3264 × 4896, and the resolution is reduced to 408 × 612 through gaussian blurring processing, so that unnecessary noise is removed while the integrity of image information is maintained, and the image resolution is reduced, thereby accelerating the speed of flash analysis and detection, as shown in fig. 4.
(3.1) over-dividing the single-frame image to obtain n initial regions r1,r2,...,rnForming an initial region set R ═ { R ═ R1,r2,...,rn}; at the same time, an initialization similarity set is constructedS，
Figure BDA0001603773750000061
(3.2) calculating the similarity s (r) of every two adjacent areasi,rj) The similarity index adopts color similarity, size similarity, texture similarity, coincidence similarity and four complementary similarities, and the four complementary similarities are respectively combined for the areas from different aspects;
(a) the color similarity is calculated as follows:
dividing each initial region into q intervals according to pixel values to obtain a normalized color histogram
Figure BDA0001603773750000062
Wherein the content of the first and second substances,
Figure BDA0001603773750000063
a normalization frequency count representing a kth pixel value interval in the ith initial region;
two adjacent regions ri,rjThe color similarity of (a) is:
Figure BDA0001603773750000064
merge ri,rjThe new region r obtainedijThe color histogram is:
Figure BDA0001603773750000065
size(rij)＝size(ri)+size(rj) (33)
wherein size (r)i) Is an initial region riThe size of (d);
(b) the texture similarity calculation method comprises the following steps:
calculating the SIFT (scale-invariant feature transform) feature of each initial region, wherein in the invention, the value of a scale factor in the patent is 1, and the width of a histogram interval is 10 to obtain
Figure BDA0001603773750000066
Obtained as a textural feature.
Wherein the content of the first and second substances,
Figure BDA0001603773750000067
representing the kth texture feature in the ith initial region;
two adjacent regions ri,rjThe texture similarity of (a) is:
Figure BDA0001603773750000071
merge ri,rjThe new region r obtainedijThe histogram of the texture is
Figure BDA0001603773750000072
(C) The calculation method of the size similarity comprises the following steps:
Figure BDA0001603773750000073
where size (im) represents the size of the entire grayscale image.
(d) The calculation method of the coincidence similarity comprises the following steps:
Figure BDA0001603773750000074
wherein BBijTo surround only the region riAnd rjThe bounding box of (1).
And finally, integrating the four similarities to obtain the final similarity:
s(ri,rj)＝a1Scolor(ri,rj)+a2Stexture(ri,rj)+a3Ssize(ri,rj)+a4Sfill(ri,rj)(38)
wherein a isi∈{0,1}，Indicating whether the corresponding similarity is used.
(3.3) all the similarities form a similarity set S, and the two adjacent regions S (r) with the highest similarity in the similarity seti,rj) Max(s), merge the two regions rij＝ri∪rjCalculating the new region rijAnd (3) updating the similarity set with the similarity of adjacent regions: deleting the region r with which the data is mergediAll similarities s (r) of the correlationi,r*) And with the region r being mergedjAll similarities s (r) of the correlationj,r#) And adding the updated similarity set as follows: adding a new region rijSimilarity to adjacent regions;
and (3.4) circulating the step 3.2-3.3, and ending the circulation when the similarity set S is an empty set. Completing the combination of the initial regions to obtain a candidate region set R;
(3.5) screening an upper chip area and a lower chip area to be analyzed from the candidate area set R, as shown in FIG. 6, so that the size of the area to be analyzed is in the range of one third to one half of the size of the original packaging piece, and the two areas to be analyzed need to be respectively contained
Figure BDA0001603773750000075
And
Figure BDA0001603773750000076
the resolution of the 4 key geometric points and the packaging original sheet is a x b.
(3.4) finishing the circulation to obtain a candidate region R; the selective search algorithm step described above is performed on the gray-scale image shown in fig. 4, and 214 candidate regions are obtained, wherein a part of the candidate regions are shown in fig. 5.
(3.5) regional screening: and setting a screening condition, and screening a chip region to be analyzed from the candidate region R. The screening of the grayscale image shown in fig. 4 will result in the region to be analyzed as shown in the right side images of fig. 6.
(4) Performing threshold-based image segmentation on the screened region to be analyzed, wherein the step is realized by the following substeps:
(4) performing threshold-based image segmentation on the screened region to be analyzed, wherein the step is realized by the following substeps:
(4.1) calculating a normalized histogram of the region to be analyzed. Let {0, 1.,. L-1} denote L different gray levels in a region image of size M N pixels, NzIndicating the number of pixels with a gray level z. The total number of pixels in the area image is MN ═ n0+n1+n2+...+nL-1，pz＝nzthe/MN indicates a normalization frequency of z, which is a gray level in the histogram, and is 0,1, 2.
(4.2) for u-0, 1,21(u), the calculation formula is as follows:
Figure BDA0001603773750000081
(4.3) for u 0,1, 2.., L-1, calculate the cumulative mean m (u) as follows:
Figure BDA0001603773750000082
(4.4) calculating the Global Gray mean mGThe calculation formula is as follows:
Figure BDA0001603773750000083
(4.5) for u-0, 1,2
Figure BDA0001603773750000084
The calculation formula is as follows:
Figure BDA0001603773750000085
(4.6) obtaining the Otsu threshold u by maximizing the objective function (13)*If the maximum values are not unique, u is obtained by averaging the respective detected maximum values u*；
Figure BDA0001603773750000086
(4.7) based on the optimal threshold u*The region image is segmented by the following formula to obtain a segmented image:
Figure BDA0001603773750000091
f (x, y) is the pixel of the pixel point (x, y) in the image of the area to be analyzed. And the segmented pixels are 1 and represent the detected flash part in the production process. The black part in the dashed line frame in fig. 7 is the flash component in the chip, so that the flash situation can be visually seen, and if the flash situation is serious, the packaging process should be immediately stopped, the mold parameters should be checked, and the generation of more flashes is avoided.
It should be understood that the present invention is not limited to the specific experimental implementation and experimental conditions of the above-mentioned embodiments, and that equivalent modifications or substitutions can be made by those skilled in the art without departing from the spirit of the present invention, and the scope of the present invention is defined by the claims.

Claims (4)

1. A semiconductor chip flash rapid detection method based on an image segmentation algorithm is characterized by comprising the following steps:
(1) acquiring a multi-frame gray image of a packaging original sheet by using an industrial camera;
(2) performing Gaussian blur preprocessing on each frame of image;
(3) based on selective search image segmentation, hundreds of candidate regions are obtained and chip regions to be analyzed are screened out, and the method is realized by the following substeps:
(3.1) over-dividing the single-frame image to obtain n initial regions r1,r2,...,rnForming an initial region set R ═ { R ═ R1,r2,...,rn}; meanwhile, an initialization similarity set S is constructed,
Figure FDA0002477359020000011
(3.2) calculating the similarity s (r) of every two adjacent areasi,rj) The similarity index adopts color similarity, size similarity, texture similarity, coincidence similarity and four complementary similarities, and the four complementary similarities are respectively combined for the areas from different aspects;
(a) the color similarity is calculated as follows:
dividing each initial region into e intervals according to pixel values to obtain a normalized color histogram
Figure FDA0002477359020000012
Wherein the content of the first and second substances,
Figure FDA0002477359020000013
the normalization frequency number of the ith pixel value interval in the ith initial area is represented;
two adjacent regions ri,rjThe color similarity of (a) is:
Figure FDA0002477359020000014
merge ri,rjThe new region r obtainedijThe color histogram is:
Figure FDA0002477359020000015
size(rij)＝size(ri)+size(rj) (4)
wherein size (r)i) Is an initial region riThe size of (d);
(b) the texture similarity calculation method comprises the following steps:
calculating the SIFT (scale-invariant feature transform) feature of each initial region to obtain
Figure FDA0002477359020000016
Obtaining the number of the texture features as v; wherein the content of the first and second substances,
Figure FDA0002477359020000017
representing the ith texture feature in the ith initial region;
two adjacent regions ri,rjThe texture similarity of (a) is:
Figure FDA0002477359020000021
merge ri,rjThe new region r obtainedijThe histogram of the texture is
Figure FDA0002477359020000022
(C) The calculation method of the size similarity comprises the following steps:
Figure FDA0002477359020000023
wherein size (im) represents the size of the entire grayscale image;
(d) the calculation method of the coincidence similarity comprises the following steps:
Figure FDA0002477359020000024
wherein BBijTo surround only the region riAnd rjThe bounding box of (1);
and finally, integrating the four similarities to obtain the final similarity:
s(ri,rj)＝a1Scolor(ri,rj)+a2Stexture(ri,rj)+a3Ssize(ri,rj)+a4Sfill(ri,rj) (9)
wherein a ism∈{0,1}，m＝1,2,3,4, indicating whether the corresponding similarity is adopted;
(3.3) all the similarities form a similarity set S, and the two adjacent regions S (r) with the highest similarity in the similarity seti,rj) Max(s), merge the two regions rij＝ri∪rjCalculating a new region rijAnd (3) updating the similarity set with the similarity of adjacent regions: deleting the region r with which the data is mergediAll similarities s (r) of the correlationi,r*) And with the region r being mergedjAll similarities s (r) of the correlationj,r#) And adding the updated similarity set as follows: adding a new region rijSimilarity to adjacent regions;
(3.4) circulating the steps (3.2) - (3.3), and ending circulation when the similarity set S is an empty set; completing the combination of the initial regions to obtain a candidate region set R;
(3.5) screening an upper chip area and a lower chip area to be analyzed from the candidate area set R, so that the size of the area to be analyzed is in the range of one third to one half of the size of the original packaging piece, and the two areas to be analyzed need to be respectively contained
Figure FDA0002477359020000031
And
Figure FDA0002477359020000032
the resolution of the packaging original sheet is a x b at the 4 key geometric points;
(4) performing threshold-based image segmentation on the screened region to be analyzed, wherein the step is realized by the following substeps:
(4.1) calculating a normalized histogram of the region to be analyzed; let {0, 1.,. L-1} denote L different gray levels in a region image of size M N pixels, NzThe number of pixels having a gray level z is represented; the total number of pixels in the area image is MN ═ n0+n1+n2+...+nL-1，pz＝nzthe/MN represents a normalization frequency with z as a gray level in the histogram, where z is 0,1, 2.
(4.2) for u-0, 1,21(u), the calculation formula is as follows:
Figure FDA0002477359020000033
(4.3) for u 0,1, 2.., L-1, calculate the cumulative mean m (u) as follows:
Figure FDA0002477359020000034
(4.4) calculating the Global Gray mean mGThe calculation formula is as follows:
Figure FDA0002477359020000035
(4.5) for u-0, 1,2
Figure FDA0002477359020000036
The calculation formula is as follows:
Figure FDA0002477359020000037
(4.6) obtaining the Otsu threshold u by maximizing the objective function (13)*If the maximum values are not unique, u is obtained by averaging the respective detected maximum values u*；
Figure FDA0002477359020000038
(4.7) based on the optimal threshold u*The region image is segmented by the following formula to obtain a segmented image:
Figure FDA0002477359020000039
f (x, y) is the pixel of the pixel point (x, y) in the image of the area to be analyzed; and the segmented pixels are 1 and represent the detected flash part in the production process.
2. The method of claim 1, wherein the packaging raw sheet in step (1) is a packaging raw sheet produced by using a semiconductor packaging press FSAM120-1US and qfnqqnwb 7X7-48L T0.75.75 dies.
3. The method of claim 1, wherein the gaussian blurring step of step (2) is as follows:
(2.1) generating a gaussian operand of size (2k +1) × (2k +1) according to equation (1):
Figure FDA0002477359020000041
wherein k is an integer, wpqIs the weight of the (p, q) coordinate position in the Gaussian operand, σ is the standard deviation of the Gaussian distribution;
(2.2) calculating the reciprocal of the sum of Gaussian operand coefficients
Figure FDA0002477359020000042
Multiplying each element in the Gaussian operand by the element to obtain a new Gaussian operand;
and (2.3) reading pixels from the image and performing convolution operation on the pixels and the Gaussian operand of the second step, thereby completing the Gaussian blur preprocessing.
4. The method according to claim 3, wherein in step (2.1), the larger σ is, the wider the frequency band of Gaussian blur is, the better the smoothing degree of the image is, and the higher the detail loss degree is; by adjusting the sigma parameter, the cancellation of image noise and preservation of integrity can be balanced.
CN201810235175.9A 2018-03-21 2018-03-21 Image segmentation algorithm-based semiconductor chip flash rapid detection method Active CN108596875B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810235175.9A CN108596875B (en) 2018-03-21 2018-03-21 Image segmentation algorithm-based semiconductor chip flash rapid detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810235175.9A CN108596875B (en) 2018-03-21 2018-03-21 Image segmentation algorithm-based semiconductor chip flash rapid detection method

Publications (2)

Publication Number Publication Date
CN108596875A CN108596875A (en) 2018-09-28
CN108596875B true CN108596875B (en) 2020-09-01

Family

ID=63627115

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810235175.9A Active CN108596875B (en) 2018-03-21 2018-03-21 Image segmentation algorithm-based semiconductor chip flash rapid detection method

Country Status (1)

Country Link
CN (1) CN108596875B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108416776B (en) * 2018-03-16 2021-04-30 京东方科技集团股份有限公司 Image recognition method, image recognition apparatus, computer product, and readable storage medium
CN110211112A (en) * 2019-05-31 2019-09-06 华中科技大学 A kind of casting defect inspection method based on filtering selective search
CN111495703B (en) * 2020-04-29 2021-05-04 京东方科技集团股份有限公司 Detection method and detection device for display panel
CN111896070A (en) * 2020-07-28 2020-11-06 三一汽车制造有限公司 Flash detection device, asphalt stirring equipment and flash detection method
CN112070090A (en) * 2020-07-31 2020-12-11 南京卫泽科技信息有限公司 Mixed target extraction method based on BPT node analysis in natural scene

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101251496A (en) * 2007-02-19 2008-08-27 奥林巴斯株式会社 Defect detection device and defect detection method
CN106228545A (en) * 2016-07-14 2016-12-14 西北民族大学 A kind of automatic division method of figure of buddha class Tangka's concordance damaged area
CN106960423A (en) * 2017-03-20 2017-07-18 浙江大学 A kind of flash detection method of Platform for IC Package

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101251496A (en) * 2007-02-19 2008-08-27 奥林巴斯株式会社 Defect detection device and defect detection method
CN106228545A (en) * 2016-07-14 2016-12-14 西北民族大学 A kind of automatic division method of figure of buddha class Tangka's concordance damaged area
CN106960423A (en) * 2017-03-20 2017-07-18 浙江大学 A kind of flash detection method of Platform for IC Package

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Detection of defects in digital texture images using segmentation;K.N.Sivabalan et.al;《 International Journal of Engineering Science and Technology》;20101231;第5187-5191页 *
Felzenszwalb P F et al..Efficient graph-based image segmentation.《International journal of computer vision》.2004, *

Also Published As

Publication number Publication date
CN108596875A (en) 2018-09-28

Similar Documents

Publication Publication Date Title
CN108596875B (en) Image segmentation algorithm-based semiconductor chip flash rapid detection method
CN106875381B (en) Mobile phone shell defect detection method based on deep learning
CN109816652B (en) Complex casting defect identification method based on gray level significance
CN112365494B (en) Ore material image segmentation method based on deep learning prediction edge
CN109829903B (en) Chip surface defect detection method based on convolution denoising autoencoder
CN115082485A (en) Method and system for detecting bubble defects on surface of injection molding product
CN114387233A (en) Sand mold defect detection method and system based on machine vision
CN115100191B (en) Metal casting defect identification method based on industrial detection
CN111415376B (en) Automobile glass subpixel contour extraction method and automobile glass detection method
CN115294410B (en) Plastic product molding control method based on pattern recognition
CN114998198A (en) Injection molding surface defect identification method
CN115880699B (en) Food packaging bag detection method and system
Yang et al. An efficient method for vehicle model identification via logo recognition
Li et al. Application of multi-scale feature fusion and deep learning in detection of steel strip surface defect
CN106056078B (en) Crowd density estimation method based on multi-feature regression type ensemble learning
CN117094975A (en) Method and device for detecting surface defects of steel and electronic equipment
CN113592782A (en) Method and system for extracting X-ray image defects of composite carbon fiber core rod
CN111868783B (en) Region merging image segmentation algorithm based on boundary extraction
CN116653457B (en) Production control method and system of digital dermatoglyph inkjet equipment
CN108717699B (en) Ultrasonic image segmentation method based on continuous minimum segmentation
CN110827309A (en) Polaroid appearance defect segmentation method based on super-pixels
CN113763404B (en) Foam image segmentation method based on optimization mark and edge constraint watershed algorithm
CN115619799A (en) Grain image segmentation method and system based on transfer learning
CN110689553B (en) Automatic segmentation method of RGB-D image
CN114354631A (en) Valve blank surface defect detection method based on vision

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
GR01 Patent grant
GR01 Patent grant