CN107316287A - A kind of defect identification method in rectangle ferrite magnetic disk sheet face - Google Patents
A kind of defect identification method in rectangle ferrite magnetic disk sheet face Download PDFInfo
- Publication number
- CN107316287A CN107316287A CN201710382603.6A CN201710382603A CN107316287A CN 107316287 A CN107316287 A CN 107316287A CN 201710382603 A CN201710382603 A CN 201710382603A CN 107316287 A CN107316287 A CN 107316287A
- Authority
- CN
- China
- Prior art keywords
- mrow
- image
- subgraph
- value
- munderover
- 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
Links
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
- G06T7/001—Industrial image inspection using an image reference approach
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Abstract
A kind of defect identification method in rectangle ferrite magnetic disk sheet face, visual apparatus obtains the image of rectangle ferrite magnetic sheet on streamline;Single rectangle ferrite magnetic sheet image is divided into by target area and background area by thresholding method;Local binarization average value processing is carried out to entire image;Calculate the standard error of the mean of binary image;It is subgraph by picture breakdown;Choose optimal threshold again to subgraph and calculate the average value after its binaryzation;Whether judge the subgraph is defect subgraph;When all zero defect images of all subgraphs, it can be determined that whole image is flawless, is otherwise exactly defective image.The present invention can be embedded into the real-time control system of rectangle ferrite magnetic sheet automated production as independent algoritic module, reach that ferrite fast and automatically changes the purpose of recognition detection, calculating speed is fast, intelligence degree is high, identification is accurate.
Description
Technical field
The present invention relates to Computer Vision Detection field, more particularly to a kind of identification of rectangle ferrite magnetic disk sheet planar defect
Method.
Background technology
Ferrite magnetic sheet is an indispensable part in various plant equipment, and industrially the demand to magnetic sheet is also got over
Come more.But these magnetic sheets can make magnetic sheet surface the defect problems such as indenture, scratch occur during production, polishing etc..
In actual production line, the defects detection to magnetic sheet is mostly to be observed by human eye with the presence or absence of defect.Manually come
Carrying out defects detection, not only efficiency comparison is low, and can not ensure the precision of detection.Therefore a kind of efficiency high, precision are developed high
Ferrite wafer surface pitting recognition methods is the active demand of current manufacturing enterprise.And it is current, with image procossing and manually
The development of intellectual technology, the product automatic detection based on computer vision is gradually developed and applied with recognition methods,
The defect of magnetic sheet can be analyzed, be recognized by image procossing.The development of computer technology causes quickly have by image
Effect, which is identified, to be possibly realized.
The content of the invention
For solve prior art to the manual identified inefficiency of rectangle ferrite magnetic sheet defect, can not accomplish high for a long time
The problem of imitating operation is based on computer there is provided a kind of surface defect recognition method of rectangle ferrite magnetic sheet, this recognition methods
Vision, can make full use of the advantage of image procossing, and classification quickly and accurately is identified to the magnetic sheet in production line, with
Coordinate the demand of production.
To achieve the above object, the present invention uses following technical scheme:
Step one, the real-time imaging of rectangle ferrite magnetic sheet on streamline is obtained by visual apparatus;
Step 2, by frame differential method, obtains single rectangle ferrite magnetic sheet image, detailed process is as follows:
2.1 calculate the difference image G of two continuous frames image1, calculation formula is:
G1=Pk-Pk-1
Wherein Pk-1For the image of previous moment, PkFor the image at current time;
2.2 pairs of difference images carry out gaussian filtering calculating, eliminate noise, and calculation formula is:
Wherein x is the pixel value of difference image;σ is the width of Gaussian function, and value takes 3,5 or 7;
2.3 calculate the integral projection of longitudinal first derivative of difference image after filtering, and calculation formula is:
Wherein Sj(x) it is integral projection value of the image under longitudinal coordinate j points, N is the height of image, and i is abscissa, M
For the width of image;
2.4 calculate the flex point of longitudinal integral projection, and calculation formula is:
H (j)=Max (Sj(x))
Wherein, H (j) is the integral projection value at flex point j positions, Max (Sj(x) it is) to ask for sequence Sj(x) peak point;
2.5 set up threshold value T, are compared with H (j), if more than the threshold value T of setting, judgement has rectangle ferrite magnetic sheet to arrive
Come, and previous moment image is stored as background image;If less than the threshold value of setting, return to step one;
Step 3, is divided into target area A and the classes of background area B two by thresholding method by image;
Step 4, threshold value T is gone out according to image p (x, y) standard deviation and mean value computationi, calculation formula is:
Ti=μz+Ci·σz
Wherein μzRepresent the average of image p (x, y) gray value, σzRepresent the standard deviation of image p (x, y) gray value, Ci
It is a control parameter, TiRepresent the threshold value calculated;
Step 5, local binarization average value processing is carried out to entire image, and embodiment is as follows:
5.1, the weight of image is first sought, image is then subjected to binary conversion treatment, calculation is:
P'(x, y)=w (x, y) p (x, y)
Wherein w (x, y) represents the weight of (x, y) this point;P (x, y) represents the gray scale after the binary conversion treatment of point (x, y)
Value;
5.2, the average gray of image after binaryzation is calculated, calculation formula is:
Wherein μ represents the average gray after binaryzation;The size of I and J representative graph elephants;
Step 6, calculates the standard error of the mean of binary image, and calculation formula is:
Wherein σaRepresent the standard error of the mean of binary image;
Step 7, judges the standard deviation calculated, and is judged as if σ is equal to zero intact above the image
Fall into, if standard deviation sigma is more than zero, following operation judges are carried out to the image;
Step 8, the K equirotal subgraph that size is M × N is decomposed into by whole image;Method for expressing is as follows:
Z=p (x, y) x ∈ 1,2 ..., I }, y ∈ 1,2 ... J }
K=I/M × J/N
Wherein M and N are the sizes of subgraph, and K is the number of the subgraph decomposed in entire image, and Z represents whole
Width image;
Step 9, optimal threshold T is chosen for subgraph again, and embodiment is as follows:
9.1, the average of each binaryzation subgraph is calculated, calculation formula is:
p'k=wk(x,y)pk(x,y)
Wherein wk(x, y) represents weight of the kth width subgraph in (x, y) this point;pk(x, y) represents that kth width subgraph exists
The gray value of (x, y) this point;p'kRepresent gray value of the kth width subgraph after the binary conversion treatment of point (x, y);Wherein μkTable
Show the average gray after kth width subgraph binaryzation;pkRepresent kth width subgraph after the binary conversion treatment of point (x, y)
Gray value;
9.2, by iteration optimal threshold system of selection, to determine the value of optimal threshold, implementation method is as follows:
Pass through a thresholding variables t (0<t<256) two of each width subgraph under each variable threshold t, are calculated
Average value after value, process above is repeated for each thresholding variables t, selects average value all less than 1
Subgraph, records the minimum gradation value in these subgraphs, and the minimum gradation value that these are obtained then is carried out into average value meter
Calculate;Calculation formula is:
Wherein J (t) represents that binaryzation average value is less than the minimal gray average value of 1 subgraph;L represents to select altogether
The number of the subgraph come, pl(x, y) represents the gray value at the l width image midpoints (x, y) selected;T represents thresholding variables;
9.3 pairs of obtained average values of minimal gray are smoothed, and calculation formula is:
Wherein H is the exponent number of moving average filter;J ' (t) is the average value after smooth;
9.4 in t and J ' under the functional relation of (t), J ' (t) is taken t values during minimum value be used as the optimal T newly produced
Value;
Step 10, according to the optimal threshold T newly produced, return to step 9.1 is recalculated flat after subgraph binaryzation
Average;
Step 11, whether according to the average value after the binaryzation for recalculating out, it is defecton to judge the subgraph
Image, and marked in whole image with original image, calculation formula is:
Wherein BkRepresent the output image after kth width image recognition;
Step 12, identification is identified to all subgraphs, when all zero defect images of all subgraphs,
It is flawless for may determine that whole image, is otherwise exactly defective image.
The present invention is handled by computer vision using the real-time imaging of rectangle ferrite magnetic sheet and is realized automatic detection;With
The position orientation triggering of ferrite wafer is realized by frame differential method, it is final to obtain clearly single rectangle ferrite magnetic sheet figure
Picture.Then first by entire image is carried out it is rough detect whether it is defective, then again to detecting defective production
Whether product carry out accurate detection identification, finally judge the product containing defective.
When calculating, on the basis of rectangle ferrite magnetic sheet real-time imaging is obtained, judged first by frame differential method
Whether there is the arrival of rectangle ferrite magnetic sheet, assign previous frame image now as background image;Threshold is carried out to the picture obtained
Value is handled, and is partitioned into the image only containing rectangle ferrite magnetic sheet.Algorithm is calculated by the average value and standard deviation of image
Optimal threshold;Then carry out binary conversion treatment is carried out to image according to optimal threshold, mark is asked to the binary image after processing
It is accurate poor, if standard deviation is zero, it is judged as no defective product, if standard deviation is not zero, divides the image into many equal portions, select again
Select optimal threshold.The average value of every width binaryzation subgraph is calculated by the optimal threshold chosen, if every width subgraph
Average value is equal to one, then is judged as no defective product, is otherwise judged as faulty goods.
The present invention can be embedded into the real-time control of rectangle ferrite magnetic sheet automated production as independent algoritic module
In system, reach that ferrite fast and automatically changes the purpose of recognition detection, calculating speed is fast, intelligence degree is high, identification is accurate.
Brief description of the drawings
Fig. 1 is workflow block diagram of the invention.
Fig. 2 contains defective picture for what the present invention was collected.
Fig. 3 is that image carries out the result after binaryzation average value processing.
Fig. 4 is to collect defective subgraph.
Fig. 5 is the result that optimal threshold T is selected.
Embodiment
Referring to the drawings, a kind of defect identification method in the rectangle ferrite magnetic disk sheet face based on computer vision, step is such as
Under:
Step one, the real-time imaging of 800 × 600 rectangle ferrite magnetic sheets on streamline is obtained by visual apparatus;
Step 2, by frame differential method, obtains single rectangle ferrite magnetic sheet image, detailed process is as follows:
2.1 calculate the difference image of two continuous frames image, and calculation formula is:
G1=Pk-Pk-1
Wherein Pk-1For the image of previous moment, PkFor the image at current time;
2.2 pairs of difference images carry out gaussian filtering calculating, eliminate noise, and the width for setting Gaussian function is 3, calculate public
Formula is:
Wherein x is the pixel value of difference image.
2.3 calculate the integral projection of longitudinal first derivative of difference image after filtering, and calculation formula is:
Wherein Sj(x) it is integral projection value of the image under longitudinal coordinate j points, i is abscissa, and the width of image is 800,
Highly it is 600.
2.4 calculate the flex point of longitudinal integral projection, and calculation formula is:
H (j)=Max (Sj(x))
If corresponding j values are 60, H (j) value 6000 is tried to achieve.
2.5 given threshold parameters for 3000, H (j) value 6000 compared with given threshold parameter 3000,6000 are more than
3000, that is, judge there is the arrival of rectangle ferrite magnetic sheet, and previous moment image Pk-1Stored as background image;
Step 3, is divided into target area A and the classes of background area B two by thresholding method by image, and to a-quadrant not
Any processing is done, and the extracting section for comprising only a-quadrant p (x, y) is come out.Calculation formula is:
Wherein f (x, y) is the gray scale of original image, and the threshold value T set in formula is image that 60, p (x, y) is after handling.
Step 4, threshold value T is gone out according to image p (x, y) standard deviation and mean value computationi, calculation formula is:
Ti=μz+Ci·σz=104.1+0.8 × 8.7=111.0
Wherein μzRepresent the average of image p (x, y) gray value, σzRepresent the standard deviation of image p (x, y) gray value, control
Parameter CiIt is set as 0.8, TiRepresent the threshold value calculated.
Step 5, local binarization average value processing is carried out to entire image, and embodiment is as follows:
5.1, the weight of image is first sought, image is then subjected to binary conversion treatment, calculation is:
P'(x, y)=w (x, y) p (x, y)
Wherein w (x, y) represents the weight of (x, y) this point;P ' (x, y) represents the ash after the binary conversion treatment of point (x, y)
Angle value;
5.2, the average gray of image after binaryzation is calculated, calculation formula is:
Wherein μ represents the average gray after binaryzation;The long I of image is 600, and wide J is 400.
Step 6, calculates the standard error of the mean of binary image, and calculation formula is:
Step 7, judges the standard deviation calculated, and standard deviation sigma carries out following more than zero, then to the image
Operation judges.
Step 8, the K equirotal subgraph that size is M × N is decomposed into by whole image.Method for expressing is as follows:
Z=p (x, y) x ∈ 1,2 ..., I }, y ∈ 1,2 ... J }
K=I/M × J/N=600/60 × 400/40=100
The long M of its neutron image is 60, a width of 40, and the number K of the subgraph decomposed in entire image is 100, Z representatives
Entire image.
Step 9, it is 100 to choose optimal threshold T again for subgraph, and embodiment is as follows:
9.1, the average of each binaryzation subgraph is calculated, calculation formula is:
p'k=wk(x,y)pk(x,y)
Wherein wk(x, y) represents weight of the kth width subgraph in (x, y) this point;pk(x, y) represents that kth width subgraph exists
The gray value of (x, y) this point;pkRepresent gray value of the kth width subgraph after the binary conversion treatment of point (x, y);Wherein μkTable
Show the average gray after kth width subgraph binaryzation;pkRepresent kth width subgraph after the binary conversion treatment of point (x, y)
Gray value.
9.2, by iteration optimal threshold system of selection, to determine the value of optimal threshold, implementation method is as follows:
Pass through a thresholding variables t (0<t<256) two of each width subgraph under each variable threshold t, are calculated
Average value after value, process above is repeated for each thresholding variables t, selects average value all less than 1
Subgraph, records the minimum gradation value in these subgraphs, and the minimum gradation value that these are obtained then is carried out into average value meter
Calculate.Calculation formula is:
Wherein J (t) represents that binaryzation average value is less than the minimal gray average value of 1 subgraph, calculating obtain for
0.3625;The number L of the subgraph chosen altogether is 34, pl(x, y) represents the l width image midpoints (x, y) selected
Gray value;T represents thresholding variables.
9.3 pairs of obtained average values of minimal gray are smoothed, and calculation formula is:
The exponent number H of wherein moving average filter is set to 4;Average value J ' (t) after smooth is 0.3544.
9.4 in t and J ' under the functional relation of (t), J ' (t) is taken to value t=122 during minimum value as newly producing most
Excellent T values.
Step 10, according to the optimal threshold T=122 newly produced, return to step 9.1 recalculates subgraph binaryzation
Average value afterwards.
Step 11, whether according to the average value after the binaryzation for recalculating out, it is defecton to judge the subgraph
Image, and marked in whole image with original image, calculation formula is:
Wherein BkRepresent the output image after kth width image recognition.
Step 12, identification is identified to all subgraphs, 7 subgraphs has been obtained for defect image, so sentencing
The image break for defect image.
Claims (1)
1. a kind of defect identification method in rectangle ferrite magnetic disk sheet face, it is characterised in that comprise the following steps:
Step one, the real-time imaging of rectangle ferrite magnetic sheet on streamline is obtained by visual apparatus;
Step 2, by frame differential method, obtains single rectangle ferrite magnetic sheet image, detailed process is as follows:
2.1 calculate the difference image G of two continuous frames image1, calculation formula is:
G1=Pk-Pk-1
Wherein Pk-1For the image of previous moment, PkFor the image at current time;
2.2 pairs of difference images carry out gaussian filtering calculating, eliminate noise, and calculation formula is:
<mrow>
<msup>
<mi>G</mi>
<mn>1</mn>
</msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mfrac>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msup>
<mi>&sigma;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
</mrow>
</msup>
</mrow>
Wherein x is the pixel value of difference image;σ is the width of Gaussian function, and value takes 3,5 or 7;
2.3 calculate the integral projection of longitudinal first derivative of difference image after filtering, and calculation formula is:
<mrow>
<msub>
<mi>S</mi>
<mi>j</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
<mi>M</mi>
</munderover>
<mo>|</mo>
<msubsup>
<mi>G</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mn>1</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msubsup>
<mi>G</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
<mn>1</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mo>,</mo>
<mn>0</mn>
<mo><</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
Wherein Sj(x) it is integral projection value of the image under longitudinal coordinate j points, N is the height of image, and i is abscissa, and M is image
Width;
2.4 calculate the flex point of longitudinal integral projection, and calculation formula is:
H (j)=Max (Sj(x))
Wherein, Max (Sj(x) it is) to ask for sequence Sj(x) peak point;
2.5 set up threshold value T, are compared with H (j), if more than the threshold value T of setting, judgement has the arrival of rectangle ferrite magnetic sheet, and
Previous moment image is stored as background image;If less than the threshold value of setting, return to step one;
Step 3, is divided into target area A and the classes of background area B two by thresholding method by image;
Step 4, threshold value T is gone out according to image p (x, y) standard deviation and mean value computationi, calculation formula is:
Ti=μz+Ci·σz
Wherein μzRepresent the average of image p (x, y) gray value, σzRepresent the standard deviation of image p (x, y) gray value, CiIt is one
Control parameter;
Step 5, carries out local binarization average value processing to entire image, comprises the following steps that:
5.1, the weight of image is first sought, image is then subjected to binary conversion treatment, calculation is:
P'(x, y)=w (x, y) p (x, y)
Wherein w (x, y) represents the weight of (x, y) this point;P (x, y) represents the gray value after the binary conversion treatment of point (x, y);
5.2, the average gray of image after binaryzation is calculated, calculation formula is:
<mrow>
<mi>&mu;</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>I</mi>
<mi>J</mi>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>I</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>J</mi>
</munderover>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
The wherein size of I and J representative graphs elephant;
Step 6, calculates the standard error of the mean of binary image, and calculation formula is:
<mrow>
<msub>
<mi>&sigma;</mi>
<mi>a</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>I</mi>
<mo>&times;</mo>
<mi>J</mi>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>I</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>J</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<mi>&mu;</mi>
<mo>-</mo>
<msup>
<mi>p</mi>
<mo>&prime;</mo>
</msup>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
1
Wherein σaRepresent the standard error of the mean of binary image;
Step 7, judges the standard deviation calculated, if σaIt is judged as zero defect above the image equal to zero, such as
Fruit standard deviation sigmaaMore than zero, then following operation judges are carried out to the image;
Step 8, the K equirotal subgraph that size is M × N is decomposed into by whole image;Method for expressing is as follows:
Z=p (x, y) | x ∈ 1,2 ..., I }, y ∈ 1,2 ... J }
<mrow>
<mi>Z</mi>
<mo>=</mo>
<msubsup>
<mrow>
<mo>{</mo>
<msub>
<mi>z</mi>
<mi>k</mi>
</msub>
<mo>}</mo>
</mrow>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</msubsup>
<mo>=</mo>
<msubsup>
<mrow>
<mo>{</mo>
<msub>
<mi>p</mi>
<mi>k</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mi>x</mi>
<mo>&Element;</mo>
<mo>{</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>M</mi>
<mo>}</mo>
<mo>,</mo>
<mi>y</mi>
<mo>&Element;</mo>
<mo>{</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<mi>N</mi>
<mo>}</mo>
<mo>}</mo>
</mrow>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</msubsup>
</mrow>
K=I/M × J/N
Wherein M and N are the sizes of subgraph, and K is the number of the subgraph decomposed in entire image, and Z represents view picture figure
Picture;
Step 9, optimal threshold T is chosen for subgraph, is comprised the following steps that again:
9.1, the average of each binaryzation subgraph is calculated, calculation formula is:
p'k=wk(x,y)pk(x,y)
<mrow>
<msub>
<mi>&mu;</mi>
<mi>k</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>M</mi>
<mi>N</mi>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msubsup>
<mi>P</mi>
<mi>k</mi>
<mo>&prime;</mo>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
Wherein wk(x, y) represents weight of the kth width subgraph in (x, y) this point;pk(x, y) represent kth width subgraph (x,
Y) gray value of this point;p'kRepresent gray value of the kth width subgraph after the binary conversion treatment of point (x, y);Wherein μkRepresent
Average gray after kth width subgraph binaryzation;pkRepresent ash of the kth width subgraph after the binary conversion treatment of point (x, y)
Angle value;
9.2, by iteration optimal threshold system of selection, to determine the value of optimal threshold, step is as follows:
Pass through a thresholding variables t (0<t<256) binaryzation of each width subgraph under each variable threshold t, is calculated
Average value afterwards, process above is repeated for each thresholding variables t, selects all subgraphs that average value is less than 1
Picture, records the minimum gradation value in these subgraphs, and the minimum gradation value that these are obtained then is carried out into mean value calculation;Meter
Calculating formula is:
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>L</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>l</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>L</mi>
</munderover>
<mo>{</mo>
<munder>
<mi>min</mi>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>x</mi>
<mo>&le;</mo>
<mi>M</mi>
<mo>,</mo>
<mn>1</mn>
<mo>&le;</mo>
<mi>y</mi>
<mo>&le;</mo>
<mi>N</mi>
</mrow>
</munder>
<msub>
<mi>p</mi>
<mi>l</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>}</mo>
<mo>,</mo>
<mn>1</mn>
<mo>&le;</mo>
<mi>t</mi>
<mo>&le;</mo>
<mn>255</mn>
</mrow>
Wherein L represents the number of the subgraph chosen altogether, pl(x, y) represents the l width image midpoints (x, y) selected
Gray value;
9.3 pairs of obtained average values of minimal gray are smoothed, and calculation formula is:
<mrow>
<msup>
<mi>J</mi>
<mo>&prime;</mo>
</msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>H</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>h</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
<mi>H</mi>
</munderover>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>-</mo>
<mi>h</mi>
<mo>)</mo>
</mrow>
</mrow>
Wherein H is the exponent number of moving average filter;
9.4 in t and J ' under the functional relation of (t), J ' (t) is taken t values during minimum value be used as the optimal threshold T values newly produced;
Step 10, according to the optimal threshold T newly produced, return to step 9.1 recalculates being averaged after subgraph binaryzation
Value;
Step 11, whether according to the average value after the binaryzation for recalculating out, it is defect subgraph to judge the subgraph,
And marked in whole image with original image, calculation formula is:
Wherein BkRepresent the output image after kth width image recognition;
Step 12, identification is identified to all subgraphs, when all zero defect images of all subgraphs, can be with
It is flawless to judge whole image, is otherwise exactly defective image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710382603.6A CN107316287A (en) | 2017-05-26 | 2017-05-26 | A kind of defect identification method in rectangle ferrite magnetic disk sheet face |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710382603.6A CN107316287A (en) | 2017-05-26 | 2017-05-26 | A kind of defect identification method in rectangle ferrite magnetic disk sheet face |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107316287A true CN107316287A (en) | 2017-11-03 |
Family
ID=60183857
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710382603.6A Pending CN107316287A (en) | 2017-05-26 | 2017-05-26 | A kind of defect identification method in rectangle ferrite magnetic disk sheet face |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107316287A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108038853A (en) * | 2017-12-18 | 2018-05-15 | 浙江工业大学 | A kind of ceramic tile surface defect identification method based on convolutional neural networks and Active Learning |
CN109584231A (en) * | 2018-11-28 | 2019-04-05 | 中国兵器科学研究院宁波分院 | It is a kind of complexity in structure piston blank defect inspection method |
CN111062391A (en) * | 2019-12-25 | 2020-04-24 | 创新奇智(青岛)科技有限公司 | Initial positioning method for magnetic sheet |
CN112745114A (en) * | 2020-12-23 | 2021-05-04 | 东阳富仕特磁业有限公司 | Microwave gyromagnetic ferrite preparation method based on online detection |
CN112862800A (en) * | 2021-02-25 | 2021-05-28 | 歌尔科技有限公司 | Defect detection method and device and electronic equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070263922A1 (en) * | 2001-02-13 | 2007-11-15 | Orbotech Ltd. | Multiple optical input inspection system |
CN103413288A (en) * | 2013-08-27 | 2013-11-27 | 南京大学 | LCD general defect detecting method |
CN104835156A (en) * | 2015-05-05 | 2015-08-12 | 浙江工业大学 | Non-woven bag automatic positioning method based on computer vision |
CN105513039A (en) * | 2015-07-10 | 2016-04-20 | 中国电力科学研究院 | Charged insulator string icing bridging degree intelligent image analysis method |
US20160349046A1 (en) * | 2015-05-29 | 2016-12-01 | Canon Kabushiki Kaisha | Measuring shape of specular objects by local projection of coded patterns |
CN106651825A (en) * | 2015-11-03 | 2017-05-10 | 中国科学院沈阳计算技术研究所有限公司 | Workpiece positioning and identification method based on image segmentation |
-
2017
- 2017-05-26 CN CN201710382603.6A patent/CN107316287A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070263922A1 (en) * | 2001-02-13 | 2007-11-15 | Orbotech Ltd. | Multiple optical input inspection system |
CN103413288A (en) * | 2013-08-27 | 2013-11-27 | 南京大学 | LCD general defect detecting method |
CN104835156A (en) * | 2015-05-05 | 2015-08-12 | 浙江工业大学 | Non-woven bag automatic positioning method based on computer vision |
US20160349046A1 (en) * | 2015-05-29 | 2016-12-01 | Canon Kabushiki Kaisha | Measuring shape of specular objects by local projection of coded patterns |
CN105513039A (en) * | 2015-07-10 | 2016-04-20 | 中国电力科学研究院 | Charged insulator string icing bridging degree intelligent image analysis method |
CN106651825A (en) * | 2015-11-03 | 2017-05-10 | 中国科学院沈阳计算技术研究所有限公司 | Workpiece positioning and identification method based on image segmentation |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108038853A (en) * | 2017-12-18 | 2018-05-15 | 浙江工业大学 | A kind of ceramic tile surface defect identification method based on convolutional neural networks and Active Learning |
CN108038853B (en) * | 2017-12-18 | 2020-05-26 | 浙江工业大学 | Ceramic tile surface defect identification method based on convolutional neural network and active learning |
CN109584231A (en) * | 2018-11-28 | 2019-04-05 | 中国兵器科学研究院宁波分院 | It is a kind of complexity in structure piston blank defect inspection method |
CN109584231B (en) * | 2018-11-28 | 2022-11-29 | 中国兵器科学研究院宁波分院 | Method for detecting defects of piston blank with complex inner structure |
CN111062391A (en) * | 2019-12-25 | 2020-04-24 | 创新奇智(青岛)科技有限公司 | Initial positioning method for magnetic sheet |
CN111062391B (en) * | 2019-12-25 | 2023-09-19 | 创新奇智(青岛)科技有限公司 | Magnetic sheet initial positioning method |
CN112745114A (en) * | 2020-12-23 | 2021-05-04 | 东阳富仕特磁业有限公司 | Microwave gyromagnetic ferrite preparation method based on online detection |
CN112862800A (en) * | 2021-02-25 | 2021-05-28 | 歌尔科技有限公司 | Defect detection method and device and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107316287A (en) | A kind of defect identification method in rectangle ferrite magnetic disk sheet face | |
CN107798326B (en) | Contour vision detection method | |
US10803573B2 (en) | Method for automated detection of defects in cast wheel products | |
CN107543828B (en) | Workpiece surface defect detection method and system | |
CN110175982B (en) | Defect detection method based on target detection | |
CN108898610A (en) | A kind of object contour extraction method based on mask-RCNN | |
CN108022233A (en) | A kind of edge of work extracting method based on modified Canny operators | |
CN107451601A (en) | Moving Workpieces recognition methods based on the full convolutional network of space-time context | |
CN107886496A (en) | A kind of quality determining method of the auto-parts based on form fit | |
CN106780526A (en) | A kind of ferrite wafer alligatoring recognition methods | |
CN105913415A (en) | Image sub-pixel edge extraction method having extensive adaptability | |
CN107490582A (en) | A kind of streamline Work Piece Verification System Based | |
CN110263795B (en) | Target detection method based on implicit shape model and graph matching | |
CN109101976B (en) | Method for detecting surface defects of arc-extinguishing grid plate | |
CN104198497A (en) | Surface defect detection method based on visual saliency map and support vector machine | |
CN112465809B (en) | Mold defect detection method based on image recognition, computer readable storage medium and device | |
CN115953373B (en) | Glass defect detection method, device, electronic equipment and storage medium | |
CN113516619B (en) | Product surface flaw identification method based on image processing technology | |
CN115082477B (en) | Semiconductor wafer processing quality detection method based on light reflection removing effect | |
CN113129260B (en) | Automatic detection method and device for internal defects of lithium battery cell | |
CN111178405A (en) | Similar object identification method fusing multiple neural networks | |
CN113781413A (en) | Electrolytic capacitor positioning method based on Hough gradient method | |
CN113989196A (en) | Vision-based earphone silica gel gasket appearance defect detection method | |
CN109934817B (en) | Method for detecting malformation of external contour of fruit body | |
CN113762427B (en) | Feeding abnormity detection method in industrial automation detection scene |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171103 |