CN100574360C - A kind of preprocess method that obtains difference image - Google Patents
A kind of preprocess method that obtains difference image Download PDFInfo
- Publication number
- CN100574360C CN100574360C CNB2007101766094A CN200710176609A CN100574360C CN 100574360 C CN100574360 C CN 100574360C CN B2007101766094 A CNB2007101766094 A CN B2007101766094A CN 200710176609 A CN200710176609 A CN 200710176609A CN 100574360 C CN100574360 C CN 100574360C
- Authority
- CN
- China
- Prior art keywords
- difference
- image
- width
- cloth
- translation
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Treatment Of Fiber Materials (AREA)
Abstract
The invention provides a kind of preconditioning technique that obtains difference image, particularly a kind of digital picture of two width of cloth generation certain variations to input, progressively carrying out the recovery of image horizontal stroke, vertically tension and compression deformation, rotational deformation and translation distortion handles, realize picture registration, but finally obtain the method for two width of cloth pictures of difference.Described generation certain variations is meant that horizontal, vertical one or more in tension and compression deformation, rotational deformation and the translation distortion have only taken place image, and horizontal, vertically tension and compression deformation occurs in last.It is characterized in that importing two width of cloth images, no matter have or not relative deformation between two width of cloth images, any relative deformation is arranged, all second width of cloth image is carried out handling at the recovery of horizontal, vertically tension and compression, rotation and these three kinds of distortion of translation successively, the public part of two figure is overlapped fully, carry out image difference then, export the image that a width of cloth shows out two different positions of the public part of photo especially.Adopt that this method just can be compared automatically to the photo of input, recovery and difference, obtain difference image clearly.
Description
Technical field
The present invention relates to a kind of preconditioning technique that obtains difference image, particularly a kind of digital picture of two width of cloth generation certain variations to input, progressively carrying out the recovery of image horizontal stroke, vertically tension and compression deformation, rotational deformation and translation distortion handles, realize picture registration, but finally obtain the method for two width of cloth pictures of difference.Described generation certain variations is meant that horizontal, vertical one or more in tension and compression deformation, rotational deformation and the translation distortion have only taken place image, and horizontal, vertically tension and compression deformation occurs in last.Adopt this method, can carry out automatic comparison, the recovery of image fast, carry out pretreated picture by the method and can obtain difference image clearly through difference.When needs are repeatedly taken pictures to same object, often because the variation of unavoidable objective condition in taking pictures, and make image on the picture that each time take pictures distortion occur, phenomenon such as move, therefore, can't find out changing of the object of being taken pictures itself by the image comparison.To this, can use the present invention effectively to recover to handle.
Background technology
The digital picture differential technique is widely used in being used to realize functions such as motion estimation, defects detection and image recognition among the fields such as medical science, manufacturing industry and electronic industry.To carry out image difference to digital picture, prerequisite is that the coordinate of corresponding pixel points on two width of cloth images is identical, reason owing to the operator, or the reason of imaging etc., make the image on two photos be difficult to not deform and overlap fully, this has just caused difference obstacle in realization, as if carrying out image difference, with regard to must first distortion recovering to handle, realize the coincidence of image to image.Present software image differential technique for difference before the recovery of anamorphose handle, can be divided into following several situation:
1. only can carry out image difference to the image that overlaps, do not have distortion fully.
2. only be confined to image is carried out the recovery processing of translation distortion, carry out image difference then.
3. employing is carried out difference to the method (as: amplitude spectrum of Fourier transform picture etc.) that image carries out numerical computations, thereby can carry out the preliminary treatment that the picture position is adjusted.
For digital picture, the problem that above-mentioned method exists is as follows:
1. the image that can carry out difference processing has certain limitation, the image that carries out difference is had the requirement that overlaps or only take place the translation distortion with reference picture fully;
2. adopt the method for numerical computations, handle, also brought complex mathematical to calculate and the slow new problems such as problem of processing speed though avoided the reduction of anamorphose.
So,, need a kind of preconditioning technique that can compare automatically fast, discern and recover to handle multiple anamorphose automatically for the realization of digital picture difference.The distortion of common image mainly contains horizontal stroke, vertically tension and compression deformation, rotational deformation and translation distortion, the present invention proposes the pretreated method of image difference at this three classes distortion.Image also may exist the distortion of various ways such as distortion and radial distortion simultaneously, and this method is not suitable for handling the distortion of these forms.
Summary of the invention
The present invention is directed to the defective or the deficiency that exist in the existing digital picture differential technique, a kind of preprocess method that obtains difference image is provided, adopt this method, can be to the digital picture of two width of cloth generation certain variations of input, according to certain step, carry out the recovery of anamorphose fast and handle, realize picture registration, be beneficial to obtain difference image clearly.
Described generation certain variations is meant that horizontal, vertical one or more in tension and compression deformation, rotational deformation and the translation distortion have only taken place image, and horizontal, vertically tension and compression deformation occurs in last or can suppose to occur in last.Utilize this method, programme by softwares such as VisualBasic, just can be fast to digital picture compare automatically, recovery and difference.
Technical scheme of the present invention is as follows:
A kind of preprocess method that obtains difference image is characterized in that may further comprise the steps:
Step 5 is handled the rotational deformation of first transfer image acquisition, makes by reverse rotation and treats that the leg-of-mutton corresponding sides of difference triangle and reference member parallel, and obtains second transfer image acquisition that the remaining translation of a width of cloth is out of shape; Perhaps, handle the translation distortion of first transfer image acquisition, make by reverse translation and treat that the leg-of-mutton corresponding vertex coordinate points of difference triangle and reference member is identical, obtain second transfer image acquisition of the remaining rotational deformation of a width of cloth;
Step 6 at second transfer image acquisition of remaining translation distortion, treats that by reverse translation is feasible the leg-of-mutton corresponding vertex of difference triangle and reference member overlaps, and promptly realizes picture registration; At second transfer image acquisition of remaining rotational deformation, treat that by reverse rotation is feasible the leg-of-mutton corresponding sides of difference triangle and reference member overlap, and promptly realize picture registration.
The difference zone for the treatment of in the described step 2 can be an entire image or a part.
Three characteristic points in the described step 3, it distributes and treats that the size in difference zone is complementary.
Triangle in the described step 3 is an acute triangle.
For second transfer image acquisition of the remaining translation distortion that obtains in the described step 5 or second transfer image acquisition of remaining rotational deformation, after carrying out reverse translation or reverse rotation, itself and reference member image are compared, draw both public domains, and non-public domain is cut out.
Described difference image is meant that the rgb value with two width of cloth image corresponding pixel points carries out the image that obtains behind the XOR.
Technique effect of the present invention is as follows:
1. realized recovery processing through the image of certain variations.
2. realized image and difference processing easily reference picture through certain variations.
By the recovery operation of anamorphose, can carry out horizontal stroke, vertically tension and compression, rotation and translation distortion automatically successively to input picture fast and recover to handle, realize picture registration, carry out image difference then, obtain difference image.
The present invention is called reference member with the base image of difference, with respect to reference member whether relative deformation is arranged no matter treat difference image, what kind of relative deformation is arranged, all according to certain step, progressively treat the recovery processing that difference image carries out horizontal stroke, vertically tension and compression, rotation and translation, make and treat that difference image overlaps fully with the difference zone for the treatment of of reference member epigraph, can carry out difference to this part easily then, obtain difference image clearly.
Description of drawings
Fig. 1 is pending image schematic diagram, and Fig. 1 (b) has translation, rotation and horizontal stroke, vertical tension and compression deformation with respect to Fig. 1 (a).
Fig. 2 is that the reference point point is got schematic diagram, and three characteristic points that meet the homogeneity principle of correspondence are respectively arranged among Fig. 2 (a) and among Fig. 2 (b).
Fig. 3 is that tension and compression recover to handle schematic diagram, and contrast Fig. 3 (a), Fig. 3 (b) have passed through horizontal, vertical tension and compression and recovered to handle.
Fig. 4 is that two images (Fig. 4 (a) and Fig. 4 (b)) are with respect to X-axis forward angle schematic diagram.
Fig. 5 is that rotation recovers to handle schematic diagram, and contrast Fig. 5 (a), Fig. 5 (b) have passed through rotation and recovered to handle.
Fig. 6 is that schematic diagram recover is handled in translation, among Fig. 6 (a) and Fig. 6 (b) Oxford gray partly fall the zone for reduction.
Fig. 7 is that translation recovers to handle the back schematic diagram, and Fig. 7 (a) and Fig. 7 (b) have realized picture registration.
Fig. 8 is an image schematic diagram after the difference.
Fig. 9 is a circuit board reference member image schematic diagram.
Figure 10 be with Fig. 9 same circuit board treat the difference image schematic diagram.
Figure 11 is an image schematic diagram after Fig. 9 and Figure 10 process difference and the reprocessing.
Figure 12 is the anamorphose sequential schematic.
Figure 13 is that anamorphose recovers the processing sequence schematic diagram.
Figure 14 implements the pretreated difference image method flow diagram of the present invention.
Embodiment
With Visual Basic written program is the software realization mode that example is introduced the method for the invention.
1. import two width of cloth images:
Import reference member and treat pretreated image.As Fig. 1 (a) and Fig. 1 (b), Fig. 1 (a) and Fig. 1 (b) are two photos of same object different time, wherein since in taking pictures for twice inevitably objective factor cause two film,faults by specific distortion has taken place, translation, rotation and horizontal stroke, vertical tension and compression deformation have taken place in Fig. 1 (b) Fig. 1 (a) relatively.Owing to be two photos of different time, variation has taken place in the object self of being taken pictures, and an oblique line occurred as Fig. 1 (b) lower right corner simultaneously.Two width of cloth picture sizes are identical, and length is g, and width is h.(b) figure has translation, rotation and horizontal stroke, vertical tension and compression deformation with respect to (a) figure.
2. get reference point:
Get P1, Q1 and 3 of R1 and P2, Q2 and R2 3 points at Fig. 1 (a) and the last point of Fig. 1 (b) respectively, wherein P1 and P2, Q1 and Q2, R1 and R2 are the point of the same position of figure on two width of cloth images respectively in correspondence with each other.
As shown in Figure 2, point is got and is finished the back and obtain six groups of coordinate figures of putting, promptly P1 (a1, b1), Q1 (c1, d1), R1 (e1, f1) and P2 (a2, b2), Q2 (c2, d2), R2 (e2, f2).
The point of reference point is got and is noted:
1) on two width of cloth images point of each reference point get all must be accurate, promptly must more accurately get the identical point in position on the figure of two width of cloth images.
2) choosing of three reference points should be distributed in the whole difference zone for the treatment of as far as possible on every width of cloth image.
3) three reference points should constitute an acute triangle as far as possible on every width of cloth image.
3. horizontal, vertically tension and compression recovery processing:
If to Fig. 2 (b) draw, press recover to handle after, Fig. 2 (b) goes up the each point abscissa and is changed to original m doubly, ordinate is changed to original n doubly (m, n the unknown are waited to ask), P2 (a2, b2), Q2 (c2, d2), R2 (e2, f2) become P3 shown in Fig. 3 (b) (a3, b3), Q3 (c3, d3) and R3 (e3, f3), the coordinate expression formula is:
a3＝a2×m
b3＝b2×n
c3＝c2×m
d3＝d2×n
e3＝e2×m
f3＝f2×n
By P1Q1=P3Q3, P1R1=P3R3 can get two relational expressions:
(a1-c1)
2+(b1-d1)
2＝(a3-c3)
2+(b3-d3)
2
(a1-e1)
2+(b1-f1)
2＝(a3-e3)
2+(b3-f3)
2
Further abbreviation gets:
(a1-c1)
2+(b1-d1)2＝m
2×(a2-c2)
2+n
2×(b2-d2)
2
(a1-e1)
2+(b1-f1)
2＝m
2×(a2-e2)
2+n
2×(b2-f2)
2
Can obtain m and n by these two equations, promptly
And then the abscissa of Fig. 2 (b) each point and ordinate can be become original m and n respectively doubly, and generating a width of cloth new images, Fig. 3 (b) promptly finishes horizontal stroke, vertically tension and compression recover to handle, and Fig. 3 (b) length becomes g*m, and width becomes h*n.Fig. 3 (b) Fig. 3 (a) (being original reference figure) has not relatively had horizontal stroke, vertical tension and compression deformation.
4. rotation recovers to handle:
Arbitrarily choose limit P1R1 at Fig. 4 (a), establish P1R1 and with X-axis forward angle be α, on Fig. 4 (b), choose limit P3R3 with same position, establish P3R3 and X-axis forward angle is β, so have:
α＝arctan((f1-b1)/(e1-a1))
β＝arctan(((f2-b2)×n)/((e2-a2)×m))
Then make again γ=| alpha-beta |.Make Fig. 3 be rotated counterclockwise an angle γ, generate a width of cloth new images, promptly finish rotation and recover to handle.Length g1 and the h1 expression formula of Fig. 4 (b) are as follows:
g1＝g×m×cosγ+h×n×sinγ
h1＝g×m×sinγ+h×n×cosγ
Recover distortion through rotation, the P3 on Fig. 3 (b) (a3, b3) on Fig. 5 (b), become P4 (a4, b4), the coordinate expression formula is:
a4＝a2×m×cosγ+b2×n×sinγ
b4＝b2×n/cosγ+(g×m-a2×m-b2×n×tanγ)×sinγ
Through postrotational image, shown in Fig. 5 (b), Fig. 5 (b) Fig. 5 (a) (being original reference figure) relatively is out of shape without spin.
5. translation recovers to handle:
The coordinate that Fig. 6 (a) (being original reference figure) and Fig. 6 (b) go up P1 and P4 be respectively (a1, b1) and (a4, b4), the left upper apex coordinate of establishing the rectangular area that will keep on Fig. 6 (a) and Fig. 6 (b) be respectively (i1, j1) with (i2, j2):
If a1＜a4, i1=0 then, i2=a4-a1, otherwise i1=a1-a4, i2=0;
If b1＜b4, j1=0 then, j2=b4-b1; Otherwise j1=b1-b4, j2=0.
Length and the width of establishing the rectangular area that will keep on Fig. 6 (a) and Fig. 6 (b) again are respectively g2 and h2:
If (g-i1)＜(g1-i2), g2=g-i1 then, otherwise g2=g1-i2;
If (h-j1)＜(h1-j2), h2=h-j1 then, otherwise h2=h1-j2.
Respectively Fig. 6 (a) being gone up left upper apex then is (i1, j1), length and width are respectively g2 and h2, and the last left upper apex of Fig. 6 (b) is (i2, j2), length and width are respectively two rectangular areas image in addition of g2 and h2 and reduce, and promptly obtain identical two width of cloth images of the coordinate of corresponding points on the figure, as shown in Figure 7.
As shown in Figure 7, the relative Fig. 7 (a) (being original reference figure) of Fig. 7 (b) has not had the translation distortion.
6. through above step, the recovery of anamorphose is handled and is all finished, i.e. preliminary treatment before the image difference is all finished.The coordinate for the treatment of corresponding points on difference image and the reference picture is identical.
7. image difference:
Carry out reprocessings such as image difference and denoising, binaryzation.Output one width of cloth has shown the image graph 8 of Fig. 1 (a) and Fig. 1 (b) different time object self difference especially after finishing.
8. obtain the example of circuit board difference image:
Fig. 9 is a circuit board reference member image schematic diagram.Figure 10 be with Fig. 9 same circuit board treat the difference image schematic diagram.The blackspot distribution has appearred in comparison diagram 9 and Figure 10 among Figure 10.
Figure 11 is an image schematic diagram after Fig. 9 and Figure 10 process difference and the reprocessing.Blackspot from Figure 11 distributes as can be seen, and significantly not existing together of Fig. 9 and Figure 10 is positioned at the left side for the treatment of the difference zone.
Figure 12 is the anamorphose sequential schematic.Here Ji Zai anamorphose is set in proper order: translation distortion-rotational deformation-horizontal stroke, vertical tension and compression deformation.In fact, these three kinds of distortion are to occur with the effect that superposes on real image.
Figure 13 is that anamorphose recovers the processing sequence schematic diagram.Here be to set the reverse operating that recovers to handle in proper order at the anamorphose of Figure 12: (comprising the translation processing) handled in tension and compression processing-rotation processing-reduction.
Figure 14 implements the pretreated difference image method flow diagram of the present invention.Here put down in writing general flow: import two width of cloth images-get reference point-tension and compression and recover to handle (tension and compression processings)-rotation and recover to handle (rotation processing)-reduce and handle (translation recovers processing)-difference processing-smoothing processing (denoising)-binary conversion treatment-preservation image.
The invention provides a kind of can the digital picture of two width of cloth generation certain variations of input according to certain step, being carried out the recovery of anamorphose fast and handle, realize the preprocess method of the difference image of picture registration.Described generation certain variations be meant image only taken place horizontal, vertically one or more in tension and compression deformation, rotational deformation and the translation distortion and horizontal, vertically tension and compression deformation occurs in last.Adopt this method, programme by softwares such as Visual Basic, just can be fast to digital picture compare automatically, recovery and difference.
Problem solved by the invention is:
1. realized recovery processing through the image of certain variations.
2. realized image and difference processing easily reference picture through certain variations.
At first in this explanation the base image of difference is called reference member image graph 1 (a), with respect to reference member whether relative deformation is arranged no matter treat difference image (being called Fig. 1 (b)), what kind of relative deformation is arranged, all according to certain step, progressively Fig. 1 (b) is carried out the recovery processing of horizontal stroke, vertically tension and compression, rotation and translation, Fig. 1 (b) is overlapped fully with the difference administrative division map for the treatment of of reference member epigraph, then this part is carried out difference.
Enforcement of the present invention is as follows:
1. the image on Fig. 1 (b) is defined as these three kinds of translation, rotation and horizontal stroke, vertical tension and compression with respect to the distortion that reference member took place.
2. clear and definite these three kinds of distortion occur in temporal precedence, three kinds of above related distortion, though be to occur on real image with the effect that superposes, in fact, they have order successively on taking place, the certain variations among the present invention is meant horizontal stroke, vertically tension and compression deformation occurs at last.And because the translation of image and the difference that rotational deformation does not have sequencing recover determining of Processing Algorithm for the ease of anamorphose, we might as well suppose that translation at first takes place, and are rotation then.
3. the reduction of these three kinds of distortion handled before the image difference and undertaken by the opposite sequence of its generation, promptly at first handle horizontal, vertical tension and compression deformation, the horizontal stroke in the removal of images distortion, vertical tension and compression effect obtain the transfer image acquisition 1 of remaining rotation of a width of cloth and translation effect; Handle rotational deformation then, eliminate the rotate effect on the transfer image acquisition 1, obtain the transfer image acquisition 2 of the remaining translation effect of a width of cloth; Handle the translation distortion at last.
On the image if certain distortion is arranged, then do handle at the recovery of this kind distortion after, this kind distortion just can be eliminated; If there is not this distortion, do handle at the recovery of this kind distortion after, image can be kept intact.So this method anamorphose recovers the thinking of Processing Algorithm: no matter whether image has relative deformation with respect to reference member, what kind of relative deformation is arranged, all according to certain step, progressively image is carried out the recovery processing of horizontal stroke, vertically tension and compression, rotation and translation, the difference zone for the treatment of of two width of cloth images is overlapped fully;
4. being initial point with the image upper left corner, is to the right the X-axis positive direction, is downwards the Y-axis positive direction, respectively sets up a coordinate system on Fig. 1 (a) and Fig. 1 (b);
Anamorphose is reduced handle before, at first main points are got reference point, according to 3 principles of determining a plane, on reference member, get three points and be made as P1, Q1 and R1, on pending image, choose simultaneously at identical with the reference member figure 3 and be made as P2, Q2 and R2, obtain the coordinate figure of these six points after point is got and finished, can obtain the deformation information of image with respect to reference member, the anamorphose after just can carrying out recovers to handle operation;
6. horizontal, vertically tension and compression recover to handle: the length of obtaining line segment P1Q1 and P1R1 according to the reference point coordinate, establish again Fig. 2 is drawn, after pressing recovery to handle, the each point abscissa is changed to original m doubly on Fig. 2, ordinate is changed to doubly (m of original n, n the unknown is waited to ask), P2, Q2 and R2 become P3, Q3 and R3, after obtaining the length of P3Q3 and P3R3 P1Q1=P3Q3 is arranged, P1R1=P3R3, can obtain m and n by two relational expressions, and then can with the abscissa of Fig. 2 each point and ordinate becomes original m respectively and n doubly obtains Fig. 3, promptly finish horizontal stroke, vertically tension and compression recover to handle;
7. rotation recover to be handled: establish P1R1 and P3R3 with X-axis forward angle is respectively α and β, make γ=alpha-beta again, Fig. 3 is obtained Fig. 4 around the hour hands anglec of rotation γ that wherein is in a good mood, figure on two figure can be turned to identically on angle, P3 becomes P4 at this moment;
8. reducing (translation recovery) handles: the horizontal stroke, the ordinate that compare P1 and P4 respectively, and the length of reference member and Fig. 4 and width, obtain the public domain of two figure, and non-public domain cut out, can obtain identical two width of cloth image graph 5 and Fig. 6 of the coordinate of corresponding points on the figure, the recovery of anamorphose is handled and is promptly all finished;
9. image difference: RGB (three primary colors) value to Fig. 5 and Fig. 6 corresponding pixel points is carried out XOR, exports a differentiated image graph 7, and image difference is promptly finished;
By the recovery operation of anamorphose, can carry out horizontal stroke, vertically tension and compression, rotation and translation distortion automatically successively to input picture fast and recover to handle, realize picture registration, carry out image difference then, obtain difference image.
Claims (6)
1. preprocess method that obtains difference image is characterized in that may further comprise the steps:
Step 1 is determined to carry out two width of cloth images that difference processing maybe needs to realize picture registration, and wherein a width of cloth is as reference part image, and another width of cloth is as treating difference image;
Step 2 is determined corresponding two and is treated the difference zone from two width of cloth images;
Step 3 according to the principle of homogeneity correspondence, is treated to choose three characteristic points in the difference zone at each, is linked to be triangle separately, i.e. reference member triangle and treat the difference triangle;
Step 4 is handled horizontal stroke/vertical tension and compression deformation for the treatment of difference image, makes by reverse tension and compression and treats that the leg-of-mutton corresponding sides of difference triangle and reference member equate, obtains first transfer image acquisition of remaining rotation of a width of cloth and translation distortion;
Step 5 is handled the rotational deformation of first transfer image acquisition, makes by reverse rotation and treats that the leg-of-mutton corresponding sides of difference triangle and reference member parallel, and obtains second transfer image acquisition that the remaining translation of a width of cloth is out of shape; Perhaps, handle the translation distortion of first transfer image acquisition, make by reverse translation and treat that the leg-of-mutton corresponding vertex coordinate points of difference triangle and reference member is identical, obtain second transfer image acquisition of the remaining rotational deformation of a width of cloth;
Step 6 at second transfer image acquisition of remaining translation distortion, treats that by reverse translation is feasible the leg-of-mutton corresponding vertex of difference triangle and reference member overlaps, and promptly realizes picture registration; At second transfer image acquisition of remaining rotational deformation, treat that by reverse rotation is feasible the leg-of-mutton corresponding sides of difference triangle and reference member overlap, and promptly realize picture registration.
2. the preprocess method of acquisition difference image according to claim 1 is characterized in that: the difference zone for the treatment of in the described step 2 can be an entire image or a part.
3. the preprocess method of acquisition difference image according to claim 1 is characterized in that: three characteristic points in the described step 3, it distributes and treats that the size in difference zone is complementary.
4. the preprocess method of acquisition difference image according to claim 1 is characterized in that: the triangle in the described step 3 is an acute triangle.
5. the preprocess method of acquisition difference image according to claim 1, it is characterized in that: for second transfer image acquisition of the remaining translation distortion that obtains in the described step 5 or second transfer image acquisition of remaining rotational deformation, after carrying out reverse translation or reverse rotation, itself and reference member image are compared, draw both public domains, and non-public domain is cut out.
6. the preprocess method of acquisition difference image according to claim 1 is characterized in that: described difference image is meant that the rgb value with two width of cloth image corresponding pixel points carries out the image that obtains behind the XOR.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007101766094A CN100574360C (en) | 2007-10-31 | 2007-10-31 | A kind of preprocess method that obtains difference image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007101766094A CN100574360C (en) | 2007-10-31 | 2007-10-31 | A kind of preprocess method that obtains difference image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101150651A CN101150651A (en) | 2008-03-26 |
| CN100574360C true CN100574360C (en) | 2009-12-23 |
Family
ID=39250972
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2007101766094A Expired - Fee Related CN100574360C (en) | 2007-10-31 | 2007-10-31 | A kind of preprocess method that obtains difference image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100574360C (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101551858B (en) * | 2009-05-13 | 2011-05-04 | 北京航空航天大学 | Target recognition method based on differential code and differential code mode |
| US9646191B2 (en) * | 2015-09-23 | 2017-05-09 | Intermec Technologies Corporation | Evaluating images |
| CN108540782B (en) * | 2017-03-03 | 2020-09-22 | 株式会社理光 | Method and device for acquiring difference image |
| CN113658176B (en) * | 2021-09-07 | 2023-11-07 | 重庆科技学院 | Ceramic tile surface defect detection method based on interaction attention and convolutional neural network |
-
2007
- 2007-10-31 CN CNB2007101766094A patent/CN100574360C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN101150651A (en) | 2008-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108257095B (en) | System for processing images | |
| CN102843497B (en) | Method for merging the regions in the image/video | |
| Paglieroni | Distance transforms: Properties and machine vision applications | |
| Fang et al. | Multilevel edge features guided network for image denoising | |
| CN104376548B (en) | A kind of quick joining method of image based on modified SURF algorithm | |
| CN100574360C (en) | A kind of preprocess method that obtains difference image | |
| CN105761233A (en) | FPGA-based real-time panoramic image mosaic method | |
| CN110111330B (en) | Mobile phone screen detection method | |
| US20230252605A1 (en) | Method and system for a high-frequency attention network for efficient single image super-resolution | |
| CN106169177A (en) | Image buffing method and device and electronic equipment | |
| CN113962905B (en) | Single image rain removing method based on multi-stage characteristic complementary network | |
| CN108154477B (en) | Image rotation method based on FPGA | |
| Yang et al. | Doing more with Moiré pattern detection in digital photos | |
| CN110427944A (en) | Acquisition methods, device, equipment and the storage medium of car plate detection data | |
| Chang et al. | Bilayer Markov random field method for detecting defects in patterned fabric | |
| Du et al. | Mixed high-order non-local attention network for single image super-resolution | |
| CN102938143A (en) | Image sequence ordering method | |
| Mou et al. | Synergic feature attention for image restoration | |
| Gapon et al. | Visible watermarks inpainting using deep learning for multimedia application | |
| Aksakallı et al. | A novel approach for copy-move forgery detection using bilateral filtering | |
| Ramakanth et al. | Super resolution using a single image dictionary | |
| Lanchi et al. | A novel unsharp mask sharpening method in preprocessing for face recognition | |
| Yang et al. | A novel and fast connected component count algorithm based on graph theory | |
| CN114926659B (en) | Deformation target positioning algorithm based on SIFT and CM | |
| WO2023211248A1 (en) | Method and electronic device for determining optimal global attention in deep learning model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091223 Termination date: 20101031 |