CN109785316A - A kind of apparent defect inspection method of chip - Google Patents
A kind of apparent defect inspection method of chip Download PDFInfo
- Publication number
- CN109785316A CN109785316A CN201910061382.1A CN201910061382A CN109785316A CN 109785316 A CN109785316 A CN 109785316A CN 201910061382 A CN201910061382 A CN 201910061382A CN 109785316 A CN109785316 A CN 109785316A
- Authority
- CN
- China
- Prior art keywords
- pin
- centroid
- vector
- angle
- image
- 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.)
- Granted
Links
Landscapes
- Image Analysis (AREA)
Abstract
A kind of apparent defect inspection method of chip, SOP chip image is shot using colorful CCD camera, the profile and centroid of chip circular mark and pin are extracted through a series of pretreatment, calculate the improvement environmental characteristic vector of circular mark and each centroid of pin, then matching positioning is carried out with template image, calculate affine transformation matrix, by image affine transformation template image coordinate system, finally whether the improvement environment vector determination pin of its circular mark is lacked, to its printing zone extract printed pixels and edge judge printing information whether defect, minimum circumscribed rectangle is calculated to its each pin profile and judges whether pin upwarps, it has a downwarp and crooked, oxidation is extracted to its pin field and desoldering pixel judges whether pin aoxidizes and desoldering.This method can automatic, quickly, easily and accurately judge SOP chip pin lack, upwarp, having a downwarp, crooked, desoldering and problem of oxidation, can also judge print information area whether complete display, can effectively detect SOP chip product appearance, reduce staff labor intensity.
Description
Technical field
The present invention is a kind of apparent defect inspection method of chip, belongs to Machine Vision Detection field, is a kind of based on improvement
The apparent defect inspection method of integrated antenna package (SOP) chip of environment vector quick location technique.
Background technique
In chip device production process, the apparent mass detection of chip is the essential link of one of them, at present
The SOP chip apparent mass detection of integrated circuit mainly surveys method using Manual Visual Inspection, and great work intensity be easy to cause erroneous detection, and
And detection speed and precision is lower, so that detection efficiency is low, is unable to satisfy the requirement of scope of the enterprise production, above-mentioned factor is larger
The development of China's integrated chip production industry is constrained in degree.
Currently, some intrinsic defects also limit the accuracy of SOP chip detection, such as detection speed to a certain extent
Degree and precision.
Summary of the invention
To solve the above-mentioned problems, the present invention provide one kind can it is automatic, quickly, easily and accurately judge SOP chip
Pin lacks, upwarps, having a downwarp, skew, desoldering and problem of oxidation, can also judge to print information area whether complete display, can
The effectively detection method of detection SOP chip product visual defects.The technical solution taken is a kind of apparent defects detection side of chip
Method includes the following steps:
(1) machine vision product defects detection hardware platform is built, detection chip obtains color image；
(2) image that histogram equalization is obtained by step (1), then median filter process image；
(3) image that gray processing is obtained by step (2) extracts circular mark profile and pin profile in the picture, calculates
The centroid of pin and circular mark respective profile；
(4) centroid obtained by step (3), calculate the improvement environmental characteristic of circular mark centroid and pin profile centroid to
Amount；
(5) the improvement environment vector of the centroid obtained by step (4) is similar to the improvement environment vector of template image centroid
Matching, wherein template image circular mark and pin environment vector are calculated by step (1) to step (4) in advance；
(6) affine transformation matrix is calculated by the matching centroid point obtained by step (5), then affine transformation is by step
(2) color image obtained and the pin profile extracted by step (3) are to reference picture coordinate system；
(7) the improvement environmental characteristic vector of the circular mark centroid obtained by step (4), the circular mark with template image
Centroid improves environmental characteristic vector and does similar comparison, shows pin if approximate without lacking, pin on the contrary lacks；
(8) gray processing obtains transformed image by step (6), and locating segmentation goes out to print information area, calculates font picture
Element and font edge pixel number account for the ratio of entire printing information area number of pixels, judge to print information whether defect；
(9) hsv color model transformation is obtained the color image after affine transformation by step (6), calculates each pin area
Domain defect pixel number accounts for the ratio of entire pin field number of pixels, thus judge the pin whether desoldering and oxidation；
(10) length-width ratio and horizontal tilt of the transformed pin profile minimum circumscribed rectangle obtained by step (6) are calculated
Angle, can determine whether pin upwarps and have a downwarp by length-width ratio, can determine whether pin is crooked by horizontal tilt angle.
Preferably: the step (4) the following steps are included:
(4-1) calculates improvement environmental characteristic vector of the circular mark centroid under pin centroid set；
(4-2) calculate the improvement environmental characteristic of each pin centroid under center mark and other pin centroid set to
Amount；
Preferably: the step (5) the following steps are included:
The improvement environment for improving environment vector and template image pin centroid of (5-1) calculating input image pin centroid to
Similarity between amount；
The pin centroid of the input picture of most like improvement environment vector and reference picture are mutually matched by (5-2)
It is right, while directly matching reference picture circular mark centroid and input picture circular mark centroid；
Preferably: the improved environmental characteristic vector used calculate the following steps are included:
Wherein on two-dimensional surface point c in point set P={ p1, p2...pNImproved environmental characteristic vector Context calculating
Steps are as follows:
Step 1: the unit vector of calculating point c to the point set p-shaped heartCalculation formula is as follows:
Step 2: the vector of all the points on point c to point set P is calculatedWith vectorAngle
Angle={ angle1, angle2...angleN, angle indicates 0~2 π of range, and calculation formula is as follows:
Step 3: the distance Dist={ dist of all the points on point c to point set P is calculated1, dist2...distN, it calculates
Formula is as follows:
Step 4: initialization angular histogram AngleHist [l]=0, l=1 ..., L and distance distribution histogram DistHist
[l]=0, l=1 ..., L traverses all angle collection Angle and distance set Dist statistics angular histogram and distance distribution histogram,
Wherein L indicates the resolution ratio of statistics, and statistical is as follows:
Step 5: normalization angular histogram AngleHist and distance distribution histogram DistHist obtains angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector, environmental characteristic vector Context is by angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector form, normalization formula it is as follows:
More detailed detection method is described as follows:
(1) machine vision product defects detection hardware platform is built, detection chip obtains 3 channel RGB image I1；
(2) the image I that histogram equalization is obtained by step (1)1, then the median filter process image obtains I2；
(3) gray level image I2Obtain image I3, in image I3Middle extraction circular mark profile MarkContour and pin
Profile PinContourk, k=1~K, calculate circular mark and pin respective profile centroid MarkCentroid and
PinCentroidk, k=1~K, K are pin number；
Further, step (3) includes step in detailed below:
(3-1) is by 3 channel RGB image I2Be converted to gray level image I3；
(3-2) global threshold divides I3Obtain the image I containing chip pin region4, threshold value thresh selection be generally higher than
200, it is as follows that global threshold divides formula:
Wherein, I3(i, j) indicates gray level image I3In the pixel value of coordinate (i, j), I4(i, j) indicates global threshold segmentation
Image I4In the pixel value of coordinate (i, j), thresh is the threshold value that experience is chosen；
(3-3) extracts image I using a kind of contour following algorithm4Pin profile, given threshold abandon contour area about
PinContour is obtained less than normal pins area 60% and greater than 140% profile of pin areak, k=1~K, and calculate
Its centroid PinCentroidk, k=1~K, K are pin number, and centroid calculation formula is as follows:
Wherein (xk, yk) represent k-th of profile PinCentroidkCentroid, MkIndicate that k-th of profile contains total pixel
Number, (xkm, ykm) indicate profile on m-th of pixel coordinate；
(3-4) extracts the centroid namely circle of circular mark profile MarkContour and circular mark using hough-circle transform
Heart MarkCentroid；
(4) the centroid MarkCentroid and PinCentroid obtained by step (3)k, k=1~K, calculating circular mark
Improvement environmental characteristic the vector M arkContext and PinContext of centroid and pin profile centroidk, k=1~K；
Further, step (4) includes step in detailed below:
(4-1) calculates circular mark centroid MarkCentroid in set { PinCentroid1...PinCentroidKUnder
Improvement environmental characteristic vector M arkContext；
(4-2) calculates each pin centroid PinCentroidk, k=1~K set MarkCentroid,
PinCentroid1...PinCentroidk-1, PinCentroidk+1...PinCentroidKUnder improvement environmental characteristic vector
PinContextk, k=1~K；
(5) improvement environment the vector M arkContext and PinContext of the centroid obtained by step (4)k, k=1~K
With improvement environment the vector rMarkContext and rPinContext of template image centroidk, k=1~K Similarity matching will most phase
Like the pin centroid PinCentroid for improving environment vectork, the pin centroid rPinCentroid of k=1~K and template imagek,
K=1~K is mutually matched in pairs, while directly matching input picture circular mark centroid MarkCentroid and template image
Circular mark centroid rMarkCentroid, wherein template image circular mark and the centroid of pin and environment vector are in advance by step
Suddenly (1) to step (4) step calculates；
Further step (5) includes following several detailed steps:
(5-1) calculates the improvement environment vector of the pin centroid obtained by step (4)
{PinContext1...PinContextKWith the improvement environment vector of template image pin centroid
{rPinContext1...rPinContextKBetween similarity Sim, similarity formula is as follows:
Wherein Sim (p, q) indicates PinContextpAnd rPinContextqSimilarity, Sim (p, q) closer to 0 indicate
Similarity is higher, AngleVectorpAnd DistVectorpIt is environmental characteristic vector PinContextpAngle and distance component,
Similarly, AngleVectorqAnd DistVectorqIt is environmental characteristic vector rPinContextqAngle and distance component；
(5-2) is by the pin centroid PinCentroid of the input picture of most like improvement environment vectork, k=1~K and
Reference picture rPinCentroidk, k=1~K is mutually matched in pairs, while directly matching reference picture circular mark centroid
MarkCentroid and input picture circular mark centroid rMarkCentroid；
(6) affine transformation is calculated by the consistent RANSANC algorithm of the matching centroid point random sampling obtained by step (5)
Matrix T, the RGB image I that then affine transformation is obtained by step (2)2With the pin profile extracted by step (3)
PinContourk, k=1~K to reference picture coordinate system obtains image I5With pin profile tPinContourk, k=1~K；
(7) the improvement environmental characteristic vector M arkContext of the circular mark centroid obtained by step (4), with Prototype drawing
The improvement environmental characteristic vector rMarkContext of picture does similar comparison, show if approximate pin without lacking, otherwise pin
Lack, wherein making similarity calculation using formula (4), decision threshold is selected by experience；
(8) gray processing obtains transformed image I by step (6)5, locating segmentation, which goes out, prints information area ROI, calculating word
Volumetric pixel and font edge pixel number account for the ratio of entire printing information area ROI number of pixels, judge to print whether information lacks
It falls into；
Further, step (8) includes step in detailed below:
(8-1) gray processing obtains transformed image I by step (6)5, locating segmentation go out print information area ROI；
(8-2) maximum variance between clusters binaryzation prints information area ROI, printing type face number of pixels is counted, if a
The ratio that number accounts for printing information area ROI total number is lower than the 80% of normal value, then determines to print imperfect；
(8-3) edge detection algorithm (Canny algorithm) extracts printing information area ROI, counts printing type face edge pixel
Number determines to print unintelligible if the ratio that number accounts for printing information area ROI total number is lower than the 80% of normal value；
(9) hsv color model transformation is obtained the RGB image I after affine transformation by step (6)5, calculate each pin area
Domain defect pixel number accounts for the ratio of entire pin field number of pixels, thus judge the pin whether desoldering and oxidation；
Further, step (9) includes step in detailed below:
The transformation of (9-1) hsv color model obtains transformed RGB image I by step (6)5
(9-2) counts each pin field pixel tone H at 40 °~80 °, and saturation degree S is greater than 0.15 number, if
What number accounted for the pin total number 40% or more then determines pin desoldering；
(9-3) counts each pin field pixel tone H at 70 °~130 °, and number of the lightness V less than 0.97, if
Number account for the pin total number 40% or more then determine pin aoxidize；
(10) the transformed pin profile tPinContour obtained by step (6) is calculatedk, the minimum external square of k=1~K
The length-width ratio AspectRatio of shapek, k=1~K and horizontal tilt angle HorizontalAnglek, k=1~K, by length-width ratio
AspectRatiok, k=1~K can determine whether pin upwarps and have a downwarp, by horizontal tilt angle HorizontalAnglek, k
=1~K can determine whether pin is crooked.
Further, step (10) includes step in detailed below:
(10-1) calculates the transformed pin profile tPinContour obtained by step (6)k, k=1~K's is minimum outer
Connect rectangle；
The length-width ratio AspectRatio of (10-2) calculating minimum circumscribed rectanglek, k=1~K, if length-width ratio is greater than or small
In the 20% of normal value, then show that the pin upwarps or has a downwarp；
The horizontal tilt angle HorizontalAngle of (10-3) calculating minimum circumscribed rectanglek, k=1~K, if angle
More than or less than horizontal 20 °, then show pin skew；
Wherein on two-dimensional surface point c in point set P={ p1, p2...pNImproved environmental characteristic vector Context calculating
Steps are as follows:
Step 1: the unit vector of calculating point c to the point set p-shaped heartCalculation formula is as follows:
Step 2: the vector of all the points on point c to point set P is calculatedWith vectorAngle
Angle={ angle1, angle2...angleN, angle indicates 0~2 π of range, and calculation formula is as follows:
WhereinIndicate vectorWith the angle of coordinate system,Indicate vectorWith polar angle, anglen
Indicate vectorWith vectorAngle；
Step 3: the Euclidean distance Dist={ dist of all the points on point c to point set P is calculated1, dist2...distN,
Calculation formula is as follows:
Step 4: initialization angular histogram AngleHist [l]=0, l=1 ..., L and distance distribution histogram DistHist
[l]=0, l=1 ..., L traverses all angle collection Angle and distance set Dist statistics angular histogram and distance distribution histogram,
Wherein L indicates the resolution ratio of statistics, and statistical is as follows:
Step 5: normalization angular histogram AngleHist and distance distribution histogram DistHist obtains angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector, environmental characteristic vector Context is by angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector form, normalization formula it is as follows:
Detailed description of the invention
Fig. 1 is that the present invention is based on the streams for the apparent defect inspection method of SOP chip for improving environment vector quick location technique
Cheng Tu；
Fig. 2 is SOP chip defect detection system hardware structural diagram of the present invention；
Fig. 3 is the improvement environment vector of 8 pin SOP chip template image circular marks；
Fig. 4 to Figure 11 is the improvement environment vector of the 8 each pins of pin SOP chip template image；
Figure 12 is that 8 pin SOP chip input picture circular marks improve environment vector；
Figure 13 to Figure 20 is the improvement environment vector of the 8 each pins of pin SOP chip input picture；
Figure 21 is the centroid matching result of 8 pin SOP chip template images and input picture；
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples, and embodiments of the present invention include but is not limited to
The following example.
Fig. 2 is SOP chip defect detection system hardware structural diagram of the present invention.To pacify above objective chip before detection
Industry color CCD camera and light source are filled, computer controls camera shooting and conveyer belt is mobile, when chip is moved to immediately below camera
When clap coring piece.Empty soma around chip is required to disturb when detection, conveyer belt background is black, while illumination is abundant, circle mark
Note irradiation is clear, and pin brightness is relatively high, and camera resolution is clear enough.
Fig. 3 indicates that the improvement environment vector of the circular mark of 8 pin SOP chip template images, Fig. 4 to Figure 11 indicate template
8 pin centroids of image improve environmental characteristic vector.Equally, Figure 12 indicates that the circular mark of input picture improves environment vector,
Figure 13 to Figure 20 indicates that 8 pin centroids of input picture improve environmental characteristic vector.It is all completed above by step (4), each
Curve graph represents the improvement environmental characteristic vector of corresponding circular mark and pin centroid, improves the statistics of environmental characteristic vector here
Resolution ratio L is 9, and solid line distance indicates that distance feature vector DistVector, dotted line angle indicate angle character vector
AngleHist.It can intuitively show that the improvement environmental characteristic vector of template image and input picture corresponding pin is substantially similar
, show invariance of the feature with scale, rotation, translation that the algorithm extracts, while there are also certain stability.
Figure 21 is the centroid matching result of 8 pin SOP chip template images and input picture.Template image and input picture
All step (5) are arrived by step (1).In step (3), program needs to sieve the pin profile that contour following algorithm extracts
Choosing, screening technique lean on contour area, and area approximation pin area retains, and others are rejected；Hough circle extracts circular mark simultaneously
It is influenced vulnerable to parameter, needs to be screened according to circular radius size, also to position local area search circular mark, reduce meter in this way
Calculation amount reduces noise simultaneously, increases the stability of detection.In step (4), the improvement environmental characteristic of corresponding pin is to measurer
There is very high similitude (similitude by formula (4) calculate), it can be seen that template image and input picture correspond to centroid one by one
Match.
As shown in Fig. 1 to Figure 21, the SOP chip based on 8 pin of detection for improving environment vector quick location technique is specific
Example comprises the following specific steps that:
(1) hardware detection system such as Fig. 2 is built, when transmission comes with chip transportation, computer is sent out to video camera
Shooting instruction is sent, the enough clearly input picture I of chip RGB triple channel is obtained1。
(2) the image I that histogram equalization is obtained by step (1)1, then the median filter process image obtains I2；
(3) gray level image I2Obtain image I3, since the chip of detection is 8 pin chips, therefore chip pin number K is
8, in image I3Middle extraction circular mark profile MarkContour and pin profile PinContourk, k=1~8 calculate round
The centroid MarkCentroid and PinCentroid of label and pin respective profilek, k=1~8；
Further, step (3) includes step in detailed below:
(3-1) is by 3 channel RGB image I2Be converted to gray level image I3；
(3-2) divides I with formula (1) global threshold3Obtain the image I containing chip pin region4, threshold value thresh choosing
It is taken as 220；
Wherein, I3(i, j) indicates gray level image I3In the pixel value of coordinate (i, j), I4(i, j) indicates global threshold segmentation
Image I4In the pixel value of coordinate (i, j), thresh is the threshold value that experience is chosen；
(3-3) extracts image I using contour tracing method4Pin profile, abandon pin contour area be less than normally draws
Instep accumulates 60% and obtains pin profile PinContour greater than 140% profile of pin contour areak, k=1~8, and by
Formula (2) and formula (3) calculate its centroid PinCentroidk, k=1~8；
Wherein (xk, yk) represent k-th of profile PinCentroidkCentroid, MkIndicate that k-th of profile contains total pixel
Number, (xkm, ykm) indicate profile on m-th of pixel coordinate；
(3-4) extracts the centroid namely circle of circular mark profile MarkContour and circular mark using hough-circle transform
Heart MarkCentroid；
(4) the centroid MarkCentroid and PinCentroid obtained by step (3)k, k=1~8 calculate circular mark
Improvement environmental characteristic the vector M arkContext and PinContext of centroid and pin profile centroidk, k=1~8,8 pin SOP
Improvement environmental characteristic vector calculated result such as Fig. 3 to Figure 11 and Figure 12 to Figure 20 institute of chip detection example template and input picture
Show；
Further, step (4) includes step in detailed below:
(4-1) calculates circular mark centroid MarkCentroid in set { PinCentroid1...PinCentroid8Under
Improvement environmental characteristic vector M arkContext；
(4-2) calculates each pin centroid PinCentroidk, k=1~8 set MarkCentroid,
PinCentroid1...PinCentroidk-1, PinCentroidk+1...PinCentroid8Improvement environmental characteristic vector
PinContextk, k=1~8；
(5) improvement environment the vector M arkContext and PinContext of the centroid obtained by step (4)k, k=1~8
With improvement environment the vector rMarkContext and rPinContext of template image centroidk, the Similarity matching of k=1~8 will most phase
Like the pin centroid PinCentroid for improving environment vectork, the pin centroid rPinCentroid of k=1~8 and template imagek,
K=1~8 are mutually matched in pairs, while matching the circle of input picture circular mark centroid MarkCentroid and template image
Centroid rMarkCentroid is marked, 8 pin SOP chip detection example matching results are as shown in figure 21；
Further step (5) includes following several detailed steps:
(5-1) calculates the improvement environment vector of the pin centroid obtained by step (4) with formula (4)
{PinContext1...PinContext8With the improvement environment vector of template image pin centroid
{rPinContext1...rPinContext8Between similarity Sim；
Wherein Sim (p, q) indicates PinContextpAnd rPinContextqSimilarity, Sim (p, q) closer to 0 indicate
Similarity is higher, AngleVectorpAnd DistVectorpIt is environmental characteristic vector PinContextpAngle and distance component,
Similarly, AngleVectorqAnd DistVectorqIt is environmental characteristic vector rPinContextqAngle and distance component；
(5-2) is by the pin centroid PinCentroid of the input picture of most like improvement environment vectork, the He of k=1~8
Reference picture rPinCentroidk, k=1~8 are mutually matched in pairs, while directly matching reference picture circular mark centroid
RMarkCentroid and input picture circular mark centroid MarkCentroid；
(6) affine transformation is calculated by the consistent RANSANC method of the matching centroid point random sampling obtained by step (5)
Matrix T, the RGB image I that then affine transformation is obtained by step (2)2With the pin profile extracted by step (3)
PinContourk, k=1~8 arrive reference picture coordinate system, obtain I5And tPinContourk, k=1~8；
(7) the improvement environmental characteristic vector M arkContext of the circular mark centroid obtained by step (4), with Prototype drawing
The improvement environmental characteristic vector rMarkContext of picture does similar comparison, show if approximate pin without lacking, otherwise pin
Lack, wherein using formula (4) calculating formula of similarity, decision threshold is selected by experience；
(8) gray processing obtains transformed image I by (6)5, locating segmentation, which goes out, prints information area ROI, calculating font picture
Element and font edge pixel number account for the ratio of entire printing information area ROI number of pixels, judge to print information whether defect；
Further, step (8) includes step in detailed below:
(8-1) gray processing obtains transformed image I by (6)5, locating segmentation go out print information area ROI；
(8-2) maximum variance between clusters binaryzation prints information area ROI, printing type face number of pixels is counted, if a
The ratio that number accounts for printing information area ROI total number is lower than the 80% of normal value, then determines to print imperfect；
(8-3) edge detection method (Canny method) extracts printing information area ROI, counts printing type face edge pixel
Number determines to print unintelligible if the ratio that number accounts for printing information area ROI total number is lower than the 80% of normal value；
(9) hsv color model transformation is obtained the RGB image I after affine transformation by (6)5, calculate each pin field and lack
Sunken number of pixels accounts for the ratio of entire pin field number of pixels, thus judge the pin whether desoldering and oxidation；
Further, step (9) includes step in detailed below:
The transformation of (9-1) hsv color model obtains transformed RGB image I by (6)5
(9-2) counts each pin field pixel tone H at 40 °~80 °, and saturation degree S is greater than 0.15 number, if
What number accounted for the pin total number 40% or more then determines pin desoldering；
(9-3) counts each pin field pixel tone H at 70 °~130 °, and number of the lightness V less than 0.97, if
Number account for the pin total number 40% or more then determine pin aoxidize；
(10) the transformed pin profile tPinContour obtained by (6) is calculatedk, the minimum circumscribed rectangle of k=1~8
Length-width ratio AspectRatiok, k=1~8 and horizontal tilt angle HorizontalAnglek, k=1~8, by length-width ratio
AspectRatiok, k=1~8 can determine whether pin upwarps and have a downwarp, by horizontal tilt angle HorizontalAnglek, k
=1~8 can determine whether pin is crooked.
Further, step (10) includes step in detailed below:
(10-1) calculates the transformed pin profile tPinContour obtained by (6)k, the external square of minimum of k=1~8
Shape；
The length-width ratio AspectRatio of (10-2) calculating minimum circumscribed rectanglek, k=1~8, if length-width ratio is greater than or small
In the 20% of normal value, then show that the pin upwarps or has a downwarp；
The horizontal tilt angle HorizontalAngle of (10-3) calculating minimum circumscribed rectanglek, k=1~8, if angle
More than or less than horizontal 20 °, then show pin skew；
Wherein on two-dimensional surface point c in point set P={ p1, p2...pNImproved environmental characteristic vector Context calculating
Steps are as follows:
Step 1: the unit vector of calculating point c to the point set p-shaped heartCalculation formula is as follows:
Step 2: the vector of all the points on point c to point set P is calculatedWith vectorAngle
Angle={ angle1, angle2...angleN, angle indicates 0~2 π of range, and calculation formula is as follows:
WhereinIndicate vectorWith the angle of coordinate system,Indicate vectorWith polar angle, anglen
Indicate vectorWith vectorAngle；
Step 3: the Euclidean distance Dist={ dist of all the points on point c to point set P is calculated1, dist2...distN,
Calculation formula is as follows:
Step 4: initialization angular histogram AngleHist [l]=0, l=1 ..., L and distance distribution histogram DistHist
[l]=0, l=1 ..., L traverses all angle collection Angle and distance set Dist statistics angular histogram and distance distribution histogram,
Wherein L indicates the resolution ratio of statistics, and statistical is as follows:
Step 5: normalization angular histogram AngleHist and distance distribution histogram DistHist obtains angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector, environmental characteristic vector Context is by angle environmental characteristic
Vector AngleVector and apart from environment feature vector DistVector form, normalization formula it is as follows:
According to above-described embodiment, the present invention can be realized well.It is worth noting that before based on said structure design
It puts, to solve same technical problem, even if that makes in the present invention is some without substantive change or polishing, is used
Technical solution essence still as the present invention, therefore it should also be as within the scope of the present invention.
Claims (4)
1. a kind of apparent defect inspection method of chip, characterized by the following steps:
(1) machine vision product defects detection hardware platform is built, detection chip obtains color image；
(2) image that histogram equalization is obtained by step (1), then median filter process image；
(3) image that gray processing is obtained by step (2) extracts circular mark profile and pin profile in the picture, calculates pin
With the centroid of circular mark respective profile；
(4) centroid obtained by step (3) calculates the improvement environmental characteristic vector of circular mark centroid and pin profile centroid；
(5) the improvement environment vector Similarity matching of improvement the environment vector and template image centroid of the centroid obtained by step (4),
Wherein template image circular mark and pin environment vector are calculated by step (1) to step (4) in advance；
(6) affine transformation matrix is calculated by the matching centroid point obtained by step (5), then affine transformation is obtained by step (2)
The color image that takes and the pin profile extracted by step (3) are to reference picture coordinate system；
(7) the improvement environmental characteristic vector of the circular mark centroid obtained by step (4), the circular mark centroid with template image
It improves environmental characteristic vector and does similar comparison, show pin if approximate without lacking, pin on the contrary lacks；
(8) gray processing obtains transformed image by step (6), and locating segmentation goes out to print information area, calculate font pixel and
Font edge pixel number accounts for the ratio of entire printing information area number of pixels, judge to print information whether defect；
(9) hsv color model transformation is obtained the color image after affine transformation by step (6), is calculated each pin field and is lacked
Sunken number of pixels accounts for the ratio of entire pin field number of pixels, thus judge the pin whether desoldering and oxidation；
(10) length-width ratio and the horizontal tilt angle of the transformed pin profile minimum circumscribed rectangle obtained by step (6) are calculated
Degree, can determine whether pin upwarps and have a downwarp by length-width ratio, can determine whether pin is crooked by horizontal tilt angle.
2. the apparent defect inspection method of a kind of chip according to claim 1, it is characterised in that: the step (4) includes
Following steps:
(4-1) calculates improvement environmental characteristic vector of the circular mark centroid under pin centroid set；
(4-2) calculates improvement environmental characteristic vector of each pin centroid under center mark and other pin centroid set.
3. the apparent defect inspection method of a kind of chip according to claim 1, it is characterised in that: the step (5) include with
Lower step:
The improvement environment vector for improving environment vector and template image pin centroid of (5-1) calculating input image pin centroid it
Between similarity；
The pin centroid of the input picture of most like improvement environment vector and reference picture are mutually matched in pairs, together by (5-2)
When direct matching reference picture circular mark centroid and input picture circular mark centroid.
4. the apparent defect inspection method of a kind of chip according to claim 1, it is characterised in that: step (4) and step (7) institute
The improved environmental characteristic vector stated calculate the following steps are included:
Wherein on two-dimensional surface point c in point set P={ p1, p2...pNImproved environmental characteristic vector Context calculating step
It is as follows:
Step S001: the unit vector of calculating point c to the point set p-shaped heartCalculation formula is as follows:
Step S002: the vector of all the points on point c to point set P is calculatedWith vectorIncluded angle A ngle
={ angle1, angle2...angleN, angle indicates 0~2 π of range, and calculation formula is as follows:
WhereinIndicate vectorWith the angle of coordinate system,Indicate vectorWith polar angle, anglenIt indicates
VectorWith vectorAngle；
Step S003: the Euclidean distance Dist={ dist of all the points on point c to point set P is calculated1, dist2...distN, meter
It is as follows to calculate formula:
Step S004: initialization angular histogram AngleHist [l]=0, l=1 ..., L and distance distribution histogram DistHist
[l]=0, l=1 ..., L traverses all angle collection Angle and distance set Dist statistics angular histogram and distance distribution histogram,
Wherein L indicates the resolution ratio of statistics, and statistical is as follows:
Step S005: normalization angular histogram AngleHist and distance distribution histogram DistHist obtain angle environmental characteristic to
Measure AngleVector and apart from environment feature vector DistVector, environmental characteristic vector Context from angle environmental characteristic to
It measures AngleVector and is formed apart from environment feature vector DistVector, normalization formula is as follows:
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910061382.1A CN109785316B (en) | 2019-01-22 | 2019-01-22 | Method for detecting apparent defects of chip |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910061382.1A CN109785316B (en) | 2019-01-22 | 2019-01-22 | Method for detecting apparent defects of chip |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109785316A true CN109785316A (en) | 2019-05-21 |
| CN109785316B CN109785316B (en) | 2022-12-02 |
Family
ID=66501894
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910061382.1A Active CN109785316B (en) | 2019-01-22 | 2019-01-22 | Method for detecting apparent defects of chip |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109785316B (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334760A (en) * | 2019-07-01 | 2019-10-15 | 成都数之联科技有限公司 | A kind of optical component damage detecting method and system based on resUnet |
| CN110930390A (en) * | 2019-11-22 | 2020-03-27 | 郑州智利信信息技术有限公司 | Chip pin missing detection method based on semi-supervised deep learning |
| CN110992326A (en) * | 2019-11-27 | 2020-04-10 | 江苏理工学院 | QFN chip pin image rapid inclination correction method |
| CN111260621A (en) * | 2020-01-14 | 2020-06-09 | 湖南大学 | Method for positioning and identifying surface defects of printed circuit board |
| CN111307753A (en) * | 2020-02-25 | 2020-06-19 | 杨晨露 | Fiber material detection system and method based on terahertz technology |
| CN111693546A (en) * | 2020-06-16 | 2020-09-22 | 湖南大学 | Defect detection system, method and image acquisition system |
| CN112345556A (en) * | 2020-11-23 | 2021-02-09 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
| CN112348840A (en) * | 2020-10-19 | 2021-02-09 | 江苏师范大学 | QFP chip pin defect discrimination method based on pixel region growth |
| CN112419224A (en) * | 2020-07-17 | 2021-02-26 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN113289911A (en) * | 2021-06-04 | 2021-08-24 | 宁波舜宇仪器有限公司 | Defect detection method and system for polyhedral material |
| CN114061480A (en) * | 2020-08-03 | 2022-02-18 | 上海飞机制造有限公司 | Method for detecting appearance of workpiece |
| CN114882027A (en) * | 2022-07-08 | 2022-08-09 | 南通浩盛汽车科技有限公司 | Electronic equipment chip pin defect detection method and system |
| CN115266743A (en) * | 2022-05-17 | 2022-11-01 | 江苏汤谷智能科技有限公司 | System and method for evaluating chip quality under nondestructive testing |
| CN115308222A (en) * | 2022-07-11 | 2022-11-08 | 江苏汤谷智能科技有限公司 | System and method for identifying bad chip appearance based on machine vision |
| CN115423814A (en) * | 2022-11-07 | 2022-12-02 | 江西兆驰半导体有限公司 | Chip origin positioning method and device, readable storage medium and electronic equipment |
| CN115439452A (en) * | 2022-09-13 | 2022-12-06 | 杭州凯智莆电子有限公司 | Capacitance product detection and evaluation system based on data analysis |
| CN116051564A (en) * | 2023-04-02 | 2023-05-02 | 广东仁懋电子有限公司 | Chip packaging defect detection method and system |
| CN116698753A (en) * | 2023-07-25 | 2023-09-05 | 广州纳动半导体设备有限公司 | Mini-LED panel defect detection equipment and method based on machine vision |
| CN117161624A (en) * | 2023-10-18 | 2023-12-05 | 中设智能装备（苏州）有限公司 | Intelligent welding detection device and control system |
| CN117686516A (en) * | 2024-01-29 | 2024-03-12 | 江苏优众微纳半导体科技有限公司 | Automatic chip appearance defect detection system based on machine vision |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226106A (en) * | 2013-03-15 | 2013-07-31 | 江南大学 | Machine vision based bearing dust cap quality monitoring system |
| WO2018086299A1 (en) * | 2016-11-11 | 2018-05-17 | 广东电网有限责任公司清远供电局 | Image processing-based insulator defect detection method and system |
| CN108982508A (en) * | 2018-05-23 | 2018-12-11 | 江苏农林职业技术学院 | A kind of plastic-sealed body IC chip defect inspection method based on feature templates matching and deep learning |
-
2019
- 2019-01-22 CN CN201910061382.1A patent/CN109785316B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226106A (en) * | 2013-03-15 | 2013-07-31 | 江南大学 | Machine vision based bearing dust cap quality monitoring system |
| WO2018086299A1 (en) * | 2016-11-11 | 2018-05-17 | 广东电网有限责任公司清远供电局 | Image processing-based insulator defect detection method and system |
| CN108982508A (en) * | 2018-05-23 | 2018-12-11 | 江苏农林职业技术学院 | A kind of plastic-sealed body IC chip defect inspection method based on feature templates matching and deep learning |
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334760A (en) * | 2019-07-01 | 2019-10-15 | 成都数之联科技有限公司 | A kind of optical component damage detecting method and system based on resUnet |
| CN110930390B (en) * | 2019-11-22 | 2020-09-22 | 深圳市海芯微迅半导体有限公司 | Chip pin missing detection method based on semi-supervised deep learning |
| CN110930390A (en) * | 2019-11-22 | 2020-03-27 | 郑州智利信信息技术有限公司 | Chip pin missing detection method based on semi-supervised deep learning |
| CN110992326A (en) * | 2019-11-27 | 2020-04-10 | 江苏理工学院 | QFN chip pin image rapid inclination correction method |
| CN110992326B (en) * | 2019-11-27 | 2022-08-09 | 江苏理工学院 | QFN chip pin image rapid inclination correction method |
| CN111260621A (en) * | 2020-01-14 | 2020-06-09 | 湖南大学 | Method for positioning and identifying surface defects of printed circuit board |
| CN111307753A (en) * | 2020-02-25 | 2020-06-19 | 杨晨露 | Fiber material detection system and method based on terahertz technology |
| CN113447455B (en) * | 2020-02-25 | 2022-08-12 | 苏州鱼得水电气科技有限公司 | Fiber material detection system based on terahertz technology |
| CN113447455A (en) * | 2020-02-25 | 2021-09-28 | 杨晨露 | Fiber material detection system based on terahertz technology |
| CN111693546A (en) * | 2020-06-16 | 2020-09-22 | 湖南大学 | Defect detection system, method and image acquisition system |
| CN112419224A (en) * | 2020-07-17 | 2021-02-26 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN112419224B (en) * | 2020-07-17 | 2021-08-27 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN114061480B (en) * | 2020-08-03 | 2024-04-05 | 上海飞机制造有限公司 | Method for detecting appearance of workpiece |
| CN114061480A (en) * | 2020-08-03 | 2022-02-18 | 上海飞机制造有限公司 | Method for detecting appearance of workpiece |
| CN112348840A (en) * | 2020-10-19 | 2021-02-09 | 江苏师范大学 | QFP chip pin defect discrimination method based on pixel region growth |
| CN112345556B (en) * | 2020-11-23 | 2023-07-21 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
| CN112345556A (en) * | 2020-11-23 | 2021-02-09 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
| CN113289911A (en) * | 2021-06-04 | 2021-08-24 | 宁波舜宇仪器有限公司 | Defect detection method and system for polyhedral material |
| CN115266743A (en) * | 2022-05-17 | 2022-11-01 | 江苏汤谷智能科技有限公司 | System and method for evaluating chip quality under nondestructive testing |
| CN115266743B (en) * | 2022-05-17 | 2024-02-02 | 江苏汤谷智能科技有限公司 | System and method for evaluating chip quality under nondestructive testing |
| CN114882027B (en) * | 2022-07-08 | 2022-09-06 | 南通浩盛汽车科技有限公司 | Electronic equipment chip pin defect detection method and system |
| CN114882027A (en) * | 2022-07-08 | 2022-08-09 | 南通浩盛汽车科技有限公司 | Electronic equipment chip pin defect detection method and system |
| CN115308222A (en) * | 2022-07-11 | 2022-11-08 | 江苏汤谷智能科技有限公司 | System and method for identifying bad chip appearance based on machine vision |
| CN115308222B (en) * | 2022-07-11 | 2024-02-09 | 江苏汤谷智能科技有限公司 | System and method for identifying poor chip appearance based on machine vision |
| CN115439452A (en) * | 2022-09-13 | 2022-12-06 | 杭州凯智莆电子有限公司 | Capacitance product detection and evaluation system based on data analysis |
| CN115423814A (en) * | 2022-11-07 | 2022-12-02 | 江西兆驰半导体有限公司 | Chip origin positioning method and device, readable storage medium and electronic equipment |
| CN116051564A (en) * | 2023-04-02 | 2023-05-02 | 广东仁懋电子有限公司 | Chip packaging defect detection method and system |
| CN116698753A (en) * | 2023-07-25 | 2023-09-05 | 广州纳动半导体设备有限公司 | Mini-LED panel defect detection equipment and method based on machine vision |
| CN116698753B (en) * | 2023-07-25 | 2024-03-26 | 广州纳动半导体设备有限公司 | Mini-LED panel defect detection equipment and method based on machine vision |
| CN117161624A (en) * | 2023-10-18 | 2023-12-05 | 中设智能装备（苏州）有限公司 | Intelligent welding detection device and control system |
| CN117161624B (en) * | 2023-10-18 | 2024-04-26 | 中设智能装备（苏州）有限公司 | Intelligent welding detection device and control system |
| CN117686516A (en) * | 2024-01-29 | 2024-03-12 | 江苏优众微纳半导体科技有限公司 | Automatic chip appearance defect detection system based on machine vision |
| CN117686516B (en) * | 2024-01-29 | 2024-05-10 | 江苏优众微纳半导体科技有限公司 | Automatic chip appearance defect detection system based on machine vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109785316B (en) | 2022-12-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109785316A (en) | A kind of apparent defect inspection method of chip | |
| CN106407917B (en) | The retinal vessel extracting method and system distributed based on Dynamic Multi-scale | |
| CN106570511B (en) | Keyboard character defect online detection system and method based on computer vision | |
| CN107103298B (en) | Pull-up counting system and method based on image processing | |
| CN105046252B (en) | A kind of RMB prefix code recognition methods | |
| CN108491788A (en) | A kind of intelligent extract method and device for financial statement cell | |
| CN103035013B (en) | A kind of precise motion shadow detection method based on multi-feature fusion | |
| CN108319973A (en) | Citrusfruit detection method on a kind of tree | |
| CN110276386A (en) | A kind of apple grading method and system based on machine vision | |
| CN111915704A (en) | Apple hierarchical identification method based on deep learning | |
| CN103971126A (en) | Method and device for identifying traffic signs | |
| CN109271937A (en) | Athletic ground Marker Identity method and system based on image procossing | |
| CN106446894A (en) | Method for recognizing position of spherical object based on contour | |
| CN111695373B (en) | Zebra stripes positioning method, system, medium and equipment | |
| CN105205437B (en) | Side face detection method and device based on contouring head verifying | |
| CN108985337A (en) | A kind of product surface scratch detection method based on picture depth study | |
| CN106932408B (en) | A kind of paper cocooning frame pollution silk cocoon detection method based on image procossing | |
| CN108346157A (en) | It is a kind of based on Newton's Theorem object shooting image in ellipse detection method | |
| CN111161295B (en) | Dish image background stripping method | |
| CN105069816B (en) | A kind of method and system of inlet and outlet people flow rate statistical | |
| CN104517101A (en) | Game poker card recognition method based on pixel square difference matching | |
| CN113222938A (en) | Chip defect detection method and system and computer readable storage medium | |
| CN107480678A (en) | A kind of chessboard recognition methods and identifying system | |
| CN110599463A (en) | Tongue image detection and positioning algorithm based on lightweight cascade neural network | |
| CN112016563A (en) | Method for identifying authenticity of circular seal |
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 |