CN109741377A - A kind of image difference detection method - Google Patents
A kind of image difference detection method Download PDFInfo
- Publication number
- CN109741377A CN109741377A CN201811453635.1A CN201811453635A CN109741377A CN 109741377 A CN109741377 A CN 109741377A CN 201811453635 A CN201811453635 A CN 201811453635A CN 109741377 A CN109741377 A CN 109741377A
- Authority
- CN
- China
- Prior art keywords
- image
- difference
- block
- location
- preset threshold
- 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
The invention discloses a kind of image difference detection methods, comprising the following steps: S10: extracting the insensitive characteristic point of scale of image A and image B respectively；S20: the characteristic point that image A and image B are mutually matched is calculated；S30: the finishing of location of pixels difference is carried out to image A and image B；S40: the error image C of image A Yu image B are calculated；S50: the differential image C average pixel value for carrying out pane division and pane is calculated；S60: doubtful different zones are filtered out based on preset threshold k1；S70: block division is carried out to error image C, several blocks are filtered out based on preset threshold k2；S80: respectively by each image A and image B, the region for the block that corresponding S70 is filtered out executes S10-S40, obtains block error image；S90: calculating separately the average pixel value of block error image, determines diff area based on preset threshold k3.The present invention can carry out all differences of image class file efficiently, accurately, completely to detect.
Description
Technical field
The present invention relates to field of image recognition, especially a kind of image difference detection method.
Background technique
The epoch of Digital Media have derived the image of magnanimity, and every profession and trade image comparison demand is big.Traditional artificial detection
Time-consuming and accuracy rate is low.At present image comparison can be carried out by some tools.It, can be by such as editable class document
The application such as Adobe Acrobat is to compare；And for not editable class pdf document, such as pure image class document, Adobe
Acrobat then only may compare the whole difference between image, the difference-between text and geometric figure in detail differences-such as image
Recognition effect is poor.And existing OCR(optical character identification) apply, such as Baidu OCR, ABBYY, Han Wang, Tsing-Hua University's text are logical, only may be used
It identifies text, can not judge geometric figure (such as illustration) and the color difference in picture, and its Text region effect depends on
The clarity of original text, it is high to the quality requirement of image, meanwhile, even if in the case where image clearly, Text region it is correct
Rate is not also high, so difference identification accuracy is also bad.Therefore, for image class document or picture, existing design is also
It can not relatively accurate, completely judge the difference between file.
Summary of the invention
Goal of the invention of the invention is: in view of the above problems, a kind of image difference detection method is provided, with same
Shi Shixian is between character content, geometric figure image even quick, accurate, the complete identification of color difference.
The technical solution adopted by the invention is as follows:
A kind of image difference detection method, comprising the following steps:
S10: the insensitive characteristic point of scale of image A and image B is extracted respectively；
S20: the insensitive characteristic point of scale that image A and image B are mutually matched is calculated；
S30: judging with the presence or absence of location of pixels difference between image A and image B, if so, eliminating the picture between image A and image B
Plain position difference executes S40, otherwise, executes S40；
S40: it is based on the insensitive characteristic point of matched scale, calculates the error image C of image A Yu image B；
S50: differential image C is divided into several panes by pre-set dimension a, calculates separately the average pixel value of each pane；
S60: the average pixel value of each pane and preset threshold k1 are compared, and label average pixel value reaches the window of k1
Lattice region is doubtful different zones；
S70: error image C is divided into several blocks by pre-set dimension b, the size b is greater than the size a, from all areas
In block, the block that the doubtful different zones quantity for being included reaches preset threshold k2 is filtered out；
S80: respectively by image A and image B, each extracted region for corresponding to the block that S70 is filtered out goes out, and executes respectively
S10-S40 obtains block error image corresponding to several block regions filtered out from S70；
S90: calculating separately the average pixel value of each block error image obtained in S80, and it is super to filter out average pixel value
The block error image for crossing preset threshold k3, using region corresponding to all block error images filtered out as final detection
As a result.
Further, the location of pixels difference in above-mentioned S30 are as follows: character format difference or pixel-shift.This is to difference
Two environmental factors being affected are detected, it are modified, during effectively can controlling Difference test, by extraneous factor
The degree of interference.
Further, the preset threshold k1 is adjustable.Adjustment for k1 can play the effect of adjustment difference tolerance
Fruit can control the screening of pane the primary election of difference, allow higher scene for tolerance, lower k1 is arranged can
To guarantee the covering surface of testing result, i.e. smaller difference can be also detected, and vice versa.In this way, can balanced Difference test
The load of effect and Difference test operation.
Further, set preset threshold k3 is not higher than (i.e. equal or be less than) preset threshold k1.In this way, can be real
After parameter is simply provided now, the recyclable operation of method can effectively reduce detection operation difficulty and error probability；Meanwhile for
In secondary detection, block area is big compared with pane, therefore, be arranged lesser threshold value (mean pixel is screened) may make for
The screening of difference block is more accurate.
Further, in the S30, the method for adjustment location of pixels difference are as follows:
Step 1. pixel on traversing graph A and figure B on the same position simultaneously；It finds figure A and figure B while meeting following 3 items
A list List-L is recorded in these positions L by the position L of all pixels of part:
1: two pixel color difference of condition, and scheming the pixel on B is not background colour；
Condition 2: around the figure B location L in 8 pixels, containing pixel identical with the pixel color at the L of position, if this picture
Plain position is in m；
Condition 3: figure A and scheme B there are the identical position m of pixel color at the L of position；
The color of pixel on all positions recorded in List-L is become and is schemed B corresponding position one in figure A by step 2.
Sample.
By the above method, realizes based on the location of pixels discrepancy adjustment between contrast images, there is object of reference, it is ensured that
The same degree of two images after adjustment, and then effectively filter out influence of the external interference to Difference test.
Further, in the S30, the process for eliminating location of pixels difference is the location of pixels difference based on setting
Tolerance carries out operation, and the location of pixels difference tolerance is adjustable.The adjustment of difference tolerance, it is ensured that scheme
Versatility；It is configured according to different needs, and then guarantees that detection effect meets with detection calculations and be equalized.
Further, the size of the block it is horizontal and vertical be the integral multiple of the size of the pane, with accommodate
The pane of integer amount.In this manner it is ensured that each block includes the pane of integer amount, convenient for the screening to block,
The case where preventing pane from being failed to judge and (being not counted in two adjacent blocks) or repeating judgement (being included in two adjacent blocks),
Block screening is caused the case where falsely dropping, block average pixel value calculates error occur.
In conclusion by adopting the above-described technical solution, the beneficial effects of the present invention are:
1, by means of the invention it is also possible between all differences (such as character change, geometric graph image class file or picture
Shape deformation, color, typesetting etc.) it is identified, overlay content is wide, testing result accuracy is high, detection efficiency is high.
2, the design can between image because (such as character format difference, image incline location of pixels difference caused by external interference
Tiltedly or deviate caused pixel-shift etc.) be modified, the erroneous judgement of difference between causing content because of format differences etc. is avoided, is improved
The accuracy of difference identification between text.
3, secondary difference identifying schemes of the present invention can reduce external interference (such as scanning inclination, volume as far as possible
Page, fold) influence to difference identification result, improve the accuracy of difference identification.
4, method of the invention is during Difference test, adjustable difference tolerance, is ensuring to detect essence to realize
In the case that degree is met the requirements, the load of Difference test operation is effectively reduced, with balanced testing result and testing cost expense.
Detailed description of the invention
Examples of the present invention will be described by way of reference to the accompanying drawings, in which:
Fig. 1 is the flow chart of image difference detection method.
Fig. 2 is one embodiment of image pixel filling.
Specific embodiment
All features disclosed in this specification or disclosed all methods or in the process the step of, in addition to mutually exclusive
Feature and/or step other than, can combine in any way.
Any feature disclosed in this specification (including any accessory claim, abstract), unless specifically stated,
It is replaced by other equivalent or with similar purpose alternative features.That is, unless specifically stated, each feature is a series of
An example in equivalent or similar characteristics.
A kind of image difference detection method includes the following steps (for every two images: image A and image B, multiple figures
As and so on):
S10: the insensitive characteristic point of scale (hereinafter referred to as characteristic point) of image A and image B is extracted respectively.The insensitive feature of scale
Point is the local feature of image, indicates the characteristic point insensitive to rotation, scaling, brightness change；For feature point extraction
Method, can be using SIFT, SURF, HOG scheduling algorithm in existing method.
S20: the characteristic point that image A and image B are mutually matched is calculated.After being mutually matched, i.e., it is each to obtain two images
Mapping/incidence relation between pixel, the matching process of characteristic point, existing such as cross-matched, k nearest neighbor can be selected to match,
Consistency (RANSAC) scheduling algorithm is used at random.In Feature Points Matching, need to arrive pixel-map each in image B and image
In the identical space scale of A, space scale includes the attribute informations such as inclination, scaling, rotation.
S30: judging with the presence or absence of location of pixels difference between image A and image B, if so, between eliminating image A and image B
Location of pixels difference.It in reality, there is problems, i.e. the text of two width pictures is identical, but intercharacter exists because of format
Nuance (such as font weight, font size difference), image position offsets or image inclination cause pixel-shift (can not be complete
Alignment) etc. caused by location of pixels difference the part can be judged into difference section if directly subtracted each other, caused to character
The erroneous judgement of difference, therefore, in order to avoid needing to eliminate the image of Difference test because judging by accident caused by external interference to the difference of character
Between location of pixels difference.The process of location of pixels difference can be realized by existing tool-such as OpenCV- between elimination image,
It can detecte, adjust location of pixels difference between image.Common, different for font size image, can based on font size compared with
Big image carries out equal proportion extension to the pixel of the lesser image of font size, in expansion process, in addition to realizing the expansion to font size
Exhibition also achieves the synchronous extension to font weight, otherwise similarly to the diminution of character.Identical for font size, font weight is not
With image, can the character based on the thicker image of font pixel filling is carried out to the character of the thinner image of font, or
The character of character based on the thinner image of the font image thicker to font carries out pixel corrosion, and concrete implementation belongs to existing
There is the routine techniques of image procossing.For example, as shown in Fig. 2, carrying out pixel filling method to image A based on image B are as follows:
The pixel that step 1. traverses on (traversal refers to checks one by one) figure A and figure B on the same position simultaneously (arrives on usually
Under sequence from left to right)；It finds figure A and schemes B on identical any position L while meeting all pictures of following 3 conditions
A list List-L is recorded in these positions by the position of element:
1: two pixel color difference of condition, and scheming the pixel on B is not background colour；
Condition 2: around the figure B location L in 8 pixels, containing pixel identical with the pixel color at the L of position, if this picture
Plain position is in m；
Condition 3: figure A and scheme B there are the identical position m of pixel color at the L of position.
The color of pixel on all positions recorded in List-L is become and is schemed B corresponding position in figure A by step 2.
Equally.During carrying out pixel filling to image character, adjustable difference tolerance, i.e., after control adjustment, two images
The location of pixels difference of intercharacter.
It, can be by manually distinguishing, for can not directly judge the thin of font weight or font size for apparent difference
Elementary errors is different, can measure, adjust and diversity judgement to the character of image by character measuring tool (such as OpenCV), belongs to
In the prior art, it is also not belonging to the improvement that the present invention is made, herein without being described in detail.
Adjustment process for location of pixels difference between image is to carry out operation based on setting location of pixels difference tolerance,
It can be set and adjust location of pixels difference tolerance, the value of the location of pixels difference tolerance determines to carry out location of pixels difference
After adjustment, the intercharacter same degree of two images.Permissible accuracy is higher, then difference tolerance is lower, location of pixels difference
It adjusts finer.Such as in the two images of different font sizes, the coefficient of expansion of the lesser image of escape character (ESC)；Or in difference
Extension width (the i.e. pixel shared by former character in the two images of character thickness, when the pixel for thinner image of filling character
The radius of pixel filling is carried out around position) etc., it can be used as the parameter of location of pixels difference tolerance.
S40: it is based on matched characteristic point, calculates the error image C of image A Yu image B.Based on image A and image B phase
The pixel of mutual correlation realizes the alignment between image, in the case where image alignment, by the picture of image A and each pixel of image B
Plain value correspondence is subtracted each other, and error image C can be obtained, and error image C, which can be shown, has differences place between image A and image B.
S50: differential image C is divided into several panes by pre-set dimension a, calculates separately the mean pixel of each pane
Value.
S60: the average pixel value of each pane and preset threshold k1 are compared, and label average pixel value reaches k1
Pane's area be doubtful different zones.Preset threshold k1 in different usage scenarios and different difference tolerance events in,
It can be adjusted, to reach expected testing result.
In practice, the biggish pane of average pixel value not necessarily has differences, because of external interference-such as fold, scrolling
Cause the possibility for being mistaken for doubtful different zones higher etc. the situation-for causing spatial variations big, it therefore, can't be final to this
Determine the diff area of image A Yu image B, it is also necessary to further relatively.
S70: being divided into several blocks by pre-set dimension b for error image C, from all blocks, filters out and is included
Doubtful different zones quantity reaches the block of preset threshold k2.The size of block is bigger than the size of pane, i.e., in each block
It all include a certain number of panes.The size of block is in the horizontal and vertical size integral multiple for being disposed as pane, i.e. block
Longitudinal size be pane longitudinal size integral multiple, laterally, in this way, block can accommodate the pane of integer amount.
The more block of doubtful different zones is filtered out from error image C and carries out further diversity judgement, to avoid such region
It is to be mistaken for doubtful different zones by external interference.
S80: respectively by image A and image B, each extracted region for corresponding to the block that S70 is filtered out goes out, respectively
S10-S40 is executed, block error image corresponding to several block regions filtered out from S70 is obtained.First determine
The region that the corresponding each block filtered out in the region and image B of each block filtered out is corresponded in image A, then divides
Not using the region determined in each image A as new image A, corresponding region (corresponding to same block) conduct in image B
New image A and new image B are executed S10-S40, obtain several (numbers corresponding to the block filtered out by new image B
Amount) region error image.
S90: the average pixel value of each block error image obtained in S80 is calculated separately, mean pixel is filtered out
Value be more than preset threshold k3 block error image, using block areas corresponding to all block error images filtered out as
The result of final image A and image B Difference test.Preset threshold k3 herein, can be equal with above-mentioned preset threshold k1, can also
With (preferably) smaller than preset threshold k1.
The invention is not limited to specific embodiments above-mentioned.The present invention, which expands to, any in the present specification to be disclosed
New feature or any new combination, and disclose any new method or process the step of or any new combination.
Claims (7)
1. a kind of image difference detection method, which comprises the following steps:
S10: the insensitive characteristic point of scale of image A and image B is extracted respectively；
S20: the insensitive characteristic point of scale that image A and image B are mutually matched is calculated；
S30: judging with the presence or absence of location of pixels difference between image A and image B, if so, eliminating the picture between image A and image B
Plain position difference executes S40, otherwise, executes S40；
S40: it is based on the insensitive characteristic point of matched scale, calculates the error image C of image A Yu image B；
S50: differential image C is divided into several panes by pre-set dimension a, calculates separately the average pixel value of each pane；
S60: the average pixel value of each pane and preset threshold k1 are compared, and label average pixel value reaches the window of k1
Lattice region is doubtful different zones；
S70: error image C is divided into several blocks by pre-set dimension b, the size b is greater than the size a, from all areas
In block, the block that the doubtful different zones quantity for being included reaches preset threshold k2 is filtered out；
S80: respectively by image A and image B, each extracted region for corresponding to the block that S70 is filtered out goes out, and executes respectively
S10-S40 obtains block error image corresponding to several block regions filtered out from S70；
S90: calculating separately the average pixel value of each block error image obtained in S80, and it is super to filter out average pixel value
The block error image for crossing preset threshold k3, using region corresponding to all block error images filtered out as final detection
As a result.
2. the method as described in claim 1, which is characterized in that the location of pixels difference in the S30 are as follows: character format difference
Or pixel-shift.
3. the method as described in claim 1, which is characterized in that the preset threshold k1 is adjustable.
4. method as claimed in claim 3, which is characterized in that the preset threshold k3 is not higher than the preset threshold k1.
5. the method as described in claim 1, which is characterized in that in the S30, the method for adjustment location of pixels difference are as follows:
Step 1. pixel on traversing graph A and figure B on the same position simultaneously；It finds in figure A and figure B while meeting following 3
A list List-L is recorded in these positions L by the position L of all pixels of condition:
1: two pixel color difference of condition, and scheming the pixel on B is not background colour；
Condition 2: around the figure B location L in 8 pixels, containing pixel identical with the pixel color at the L of position, if this picture
Plain position is in m；
Condition 3: figure A and scheme B there are the identical position m of pixel color at the L of position；
The color of pixel on all positions recorded in List-L is become and is schemed B corresponding position one in figure A by step 2.
Sample.
6. the method as described in one of claim 1-5, which is characterized in that in the S30, the elimination location of pixels difference
Process is the operation that the location of pixels difference tolerance based on setting carries out, and the location of pixels difference tolerance is adjustable.
7. method as claimed in claim 6, which is characterized in that the size of the block it is horizontal and vertical be the pane
Size integral multiple, to accommodate the pane of integer amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811453635.1A CN109741377B (en) | 2018-11-30 | 2018-11-30 | Image difference detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811453635.1A CN109741377B (en) | 2018-11-30 | 2018-11-30 | Image difference detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109741377A true CN109741377A (en) | 2019-05-10 |
| CN109741377B CN109741377B (en) | 2021-07-06 |
Family
ID=66358338
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811453635.1A Active CN109741377B (en) | 2018-11-30 | 2018-11-30 | Image difference detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109741377B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112669290A (en) * | 2020-12-30 | 2021-04-16 | 稿定（厦门）科技有限公司 | Image comparison method and device |
| CN113407461A (en) * | 2021-07-20 | 2021-09-17 | 北京达佳互联信息技术有限公司 | Interface test method and device, electronic equipment and storage medium |
| CN117114420A (en) * | 2023-10-17 | 2023-11-24 | 南京启泰控股集团有限公司 | Image recognition-based industrial and trade safety accident risk management and control system and method |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110181908A1 (en) * | 2010-01-26 | 2011-07-28 | Fuji Xerox Co., Ltd. | Image information processing device, computer-readable medium and image information processing method |
| CN103077515A (en) * | 2012-12-29 | 2013-05-01 | 北方工业大学 | Multi-spectral image building change detection method |
| JP2014112347A (en) * | 2012-11-08 | 2014-06-19 | Nec Corp | Image collation system, image collation method, and program |
| CN104504387A (en) * | 2014-12-16 | 2015-04-08 | 杭州华为数字技术有限公司 | Correcting method and device for text image |
| CN104751444A (en) * | 2014-12-24 | 2015-07-01 | 张�林 | Method and device for screening significant differences from similar images |
| CN105760901A (en) * | 2016-01-27 | 2016-07-13 | 南开大学 | Automatic language identification method for multilingual skew document image |
| CN106529586A (en) * | 2016-10-25 | 2017-03-22 | 天津大学 | Image classification method based on supplemented text characteristic |
| CN106845408A (en) * | 2017-01-21 | 2017-06-13 | 浙江联运知慧科技有限公司 | A kind of street refuse recognition methods under complex environment |
| CN107395919A (en) * | 2017-07-06 | 2017-11-24 | 杭州电子科技大学 | A kind of method for realizing digital printing color correction by changing characteristic file |
| WO2017206444A1 (en) * | 2016-06-03 | 2017-12-07 | 中兴通讯股份有限公司 | Method and device for detecting imaging difference, and computer storage medium |
| CN107705295A (en) * | 2017-09-14 | 2018-02-16 | 西安电子科技大学 | A kind of image difference detection method based on steadiness factor method |
-
2018
- 2018-11-30 CN CN201811453635.1A patent/CN109741377B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110181908A1 (en) * | 2010-01-26 | 2011-07-28 | Fuji Xerox Co., Ltd. | Image information processing device, computer-readable medium and image information processing method |
| JP2014112347A (en) * | 2012-11-08 | 2014-06-19 | Nec Corp | Image collation system, image collation method, and program |
| CN103077515A (en) * | 2012-12-29 | 2013-05-01 | 北方工业大学 | Multi-spectral image building change detection method |
| CN104504387A (en) * | 2014-12-16 | 2015-04-08 | 杭州华为数字技术有限公司 | Correcting method and device for text image |
| CN104751444A (en) * | 2014-12-24 | 2015-07-01 | 张�林 | Method and device for screening significant differences from similar images |
| CN105760901A (en) * | 2016-01-27 | 2016-07-13 | 南开大学 | Automatic language identification method for multilingual skew document image |
| WO2017206444A1 (en) * | 2016-06-03 | 2017-12-07 | 中兴通讯股份有限公司 | Method and device for detecting imaging difference, and computer storage medium |
| CN106529586A (en) * | 2016-10-25 | 2017-03-22 | 天津大学 | Image classification method based on supplemented text characteristic |
| CN106845408A (en) * | 2017-01-21 | 2017-06-13 | 浙江联运知慧科技有限公司 | A kind of street refuse recognition methods under complex environment |
| CN107395919A (en) * | 2017-07-06 | 2017-11-24 | 杭州电子科技大学 | A kind of method for realizing digital printing color correction by changing characteristic file |
| CN107705295A (en) * | 2017-09-14 | 2018-02-16 | 西安电子科技大学 | A kind of image difference detection method based on steadiness factor method |
Non-Patent Citations (3)
| Title |
|---|
| SEZER KARAOGLU等: "image difference correction font", 《2010 INTERNATIONAL CONFERENCE ON DIGITAL IMAGE COMPUTING: TECHNIQUES AND APPLICATIONS》 * |
| 王华等: "一种多字体印刷藏文字符的归一化方法", 《计算机应用研究》 * |
| 蔡昀哲: "基于组合不变矩和BP神经网络的数学公式符号识别", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112669290A (en) * | 2020-12-30 | 2021-04-16 | 稿定（厦门）科技有限公司 | Image comparison method and device |
| CN113407461A (en) * | 2021-07-20 | 2021-09-17 | 北京达佳互联信息技术有限公司 | Interface test method and device, electronic equipment and storage medium |
| CN113407461B (en) * | 2021-07-20 | 2024-08-13 | 北京达佳互联信息技术有限公司 | Interface testing method and device, electronic equipment and storage medium |
| CN117114420A (en) * | 2023-10-17 | 2023-11-24 | 南京启泰控股集团有限公司 | Image recognition-based industrial and trade safety accident risk management and control system and method |
| CN117114420B (en) * | 2023-10-17 | 2024-01-05 | 南京启泰控股集团有限公司 | Image recognition-based industrial and trade safety accident risk management and control system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109741377B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10565479B1 (en) | Identifying and excluding blurred areas of images of stained tissue to improve cancer scoring | |
| US8457403B2 (en) | Method of detecting and correcting digital images of books in the book spine area | |
| US20210019878A1 (en) | Image processing device, image processing method, and image processing program | |
| CN111028213A (en) | Image defect detection method and device, electronic equipment and storage medium | |
| CN109741377A (en) | A kind of image difference detection method | |
| US20090110319A1 (en) | Methods and Systems for Background Color Extrapolation | |
| CN108985305B (en) | Laser etching industrial detonator coded image positioning and correcting method | |
| CN103177249A (en) | Image processing apparatus and image processing method | |
| CN109409290B (en) | Thermometer verification reading automatic identification system and method | |
| CN110596120A (en) | Glass boundary defect detection method, device, terminal and storage medium | |
| CN109886960A (en) | The method of glass edge defects detection based on machine vision | |
| CN101299239B (en) | Method and device for acquiring character area image and character recognition system | |
| CN110399873A (en) | ID Card Image acquisition methods, device, electronic equipment and storage medium | |
| CN111340749A (en) | Image quality detection method, device, equipment and storage medium | |
| CN109978878A (en) | Color image line segment detecting method and its system based on LSD | |
| CN107292898B (en) | A kind of license plate shadow Detection and minimizing technology based on HSV | |
| CN106530273B (en) | High-precision FPC (Flexible printed Circuit) linear line detection and defect positioning method | |
| CN109858484A (en) | A kind of multi-class transformation VLP correction algorithm based on deflection evaluation | |
| JP2005283197A (en) | Detecting method and system for streak defect of screen | |
| CN106504203A (en) | A kind of Block- matching image repair method of application screening strategy | |
| CN107609453A (en) | A kind of license plate image correction, registration number character dividing method and equipment | |
| JP5424694B2 (en) | Image recognition apparatus and program | |
| US8515204B2 (en) | Correcting defects in an image | |
| CN112070771B (en) | Adaptive threshold segmentation method and device based on HS channel and storage medium | |
| CN114677670A (en) | Automatic identification and positioning method for identity card tampering |
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 |