CN102214307A - Image matching method - Google Patents
Image matching method Download PDFInfo
- Publication number
- CN102214307A CN102214307A CN 201110175096 CN201110175096A CN102214307A CN 102214307 A CN102214307 A CN 102214307A CN 201110175096 CN201110175096 CN 201110175096 CN 201110175096 A CN201110175096 A CN 201110175096A CN 102214307 A CN102214307 A CN 102214307A
- Authority
- CN
- China
- Prior art keywords
- resolution
- sample
- image
- template
- tested
- 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 an image matching method, relating to the technical field of image matching. The image matching method comprises the following steps of: S1, dynamically calculating a sampling window and sampling; S2, if the resolution of a template is same with that of the sampling window, carrying out AND-operation on a binarization image of the template and the binarized sampling image to obtain a result image, or else, directly using the binarized image as the result image; S3, calculating space distribution vectors of the result image according to a relative value vector method; S4, matching images by using a Euclidean distance method. The method disclosed by the invention can be used for detecting videos with different resolutions by using the same template and can be used for matching images at higher speed and more accuracy when the resolution of the sample to be tested is changed.
Description
Technical field
The present invention relates to technical field of image matching, relate in particular to a kind of multi-resolution adaptive image matching method.
Background technology
Images match is meant by certain matching algorithm discerns same place between two width of cloth or multiple image, as the related coefficient of the window by identical size in comparison object district and the field of search in the two dimensional image coupling, get in the field of search the maximum pairing window center point of related coefficient as same place.Its essence is under the condition of primitive similarity the best search problem of utilization matching criterior.
The common method of images match has two kinds in the digital video content identification: first kind for adopting image space distribution histogram monitoring method; Second kind for directly utilizing image pixel value to mate.
The template (promptly be used for carry out the original image of images match) of the method that adopts the monitoring of image space distribution histogram after to binaryzation carried out spatial division, count the number of white pixel in each zone of division, draw out the space distribution histogram, obtain the N dimensional vector of a reflection template space distribution by this histogram; Then sample to be tested (will carry out the image of search pattern therein) is taken a sample, the size of sampling window (being generally the rectangular area) keeps identical with position and template, after sampled picture (image that comprises in the sampling window) binaryzation, utilize identical method to calculate a vector again, utilize algorithm of support vector machine (Support Vector Machine at last, SVM) carry out classification and matching, finally find out with template be worth convergent-divergent in proportion and the two contains the sampled picture of identical content.When this method is used template space distribution histogram calculation vector, what use is the interior white pixel number in each zone, absolute pixel quantity just, like this when sample to be tested resolution and source sample (extracting used video or the image of template) are inconsistent, though can change the size of sampling window according to ratio, but because use is absolute pixel quantity, the meeting difference is very big on pixel quantity, and the effect of identification can descend greatly; And use SVM to classify, the algorithm complex height, real-time can not satisfy practical requirement.
The pixel value of the colored template that the method utilization that directly utilizes image pixel value to mate obtains directly mates, and judges by the gap of corresponding point pixel value in calculating and the sample to be tested whether template exists.This algorithm sampling window is identical with template size, and template and sampled picture are carried out individual element relatively, judges that according to pre-set threshold whether template exists.This method is used by pixel and is compared, and the sampling window size is fixing, can not adapt to the situation of change resolution equally, and computation complexity influenced by the template image size too big, can not requirement of real time.
Summary of the invention
(1) technical matters that will solve
The technical problem to be solved in the present invention is: a kind of image matching method that situation, algorithm complex are low and real-time is good that can adapt to multiresolution is provided.
(2) technical scheme
For addressing the above problem, the invention provides a kind of image matching method, the method comprising the steps of:
S1. according to the resolution of source sample and the resolution of sample to be tested, the dynamic calculation sampling window, and take a sample from described sample to be tested according to the sampling window that calculates;
S2. with the sampled picture binaryzation,, then the sampled picture after template binary image and the binaryzation is carried out and operation if template resolution is identical with sampling window resolution, remove the not point in template, obtain result images, otherwise, directly with the image as a result of of the image after the binaryzation;
S3. according to the relative value vector method, the space distribution vector of the result images that calculation procedure S2 obtains;
S4. adopt Euclidean distance method to carry out images match.
Wherein, step S1 further comprises:
S1.1 is if sample to be tested resolution is identical with the source sample resolution, and then sampling window is identical with template size, and carries out S1.4, otherwise, according to the resolution calculating resolution variation scale factor of source sample and sample to be tested, and execution in step S1.2;
S1.2 is according to described change resolution scale factor calculation sampling window;
S1.3 detects the variation of sample to be tested resolution in real time, if change, then returns step S1.1, otherwise, execution in step S1.4;
S1.4 takes a sample from sample to be tested according to sampling window.
Wherein, in step S1.1, obtain described change resolution scale factor divided by described source sample resolution with described sample to be tested resolution.
Wherein, in step S1.2, multiply by template resolution, obtain sampling window resolution with described change resolution scale factor.
Wherein, step S3 further comprises:
S3.1 carries out image segmentation to the result images that step S2 obtains;
S3.2 calculates the number of white point in each cut zone;
S3.3 calculates the number percent that the white point number accounts for the pixel value of corresponding divided areas, described number percent is calculated the space distribution vector of described result images as vector element.
Wherein, in step S4, ignoring described result images space distribution vector during coupling is the value at 0 place at described template binary image vector correspondence position.
(3) beneficial effect
Method of the present invention only uses same template to detect the video of different resolution, also can carry out when the sample to be tested change resolution faster and more accurate images match.
Description of drawings
Fig. 1 is the image matching method process flow diagram according to one embodiment of the present invention;
Fig. 2 is according to sampling flowsheet figure in the image matching method of one embodiment of the present invention;
Fig. 3 is according to the process flow diagram that calculates the space distribution vector of sampled picture in the image matching method of one embodiment of the present invention.
Embodiment
The image matching method that the present invention proposes reaches embodiment in conjunction with the accompanying drawings and is described in detail as follows.
As shown in Figure 1, the image matching method according to one embodiment of the present invention comprises step:
S1. initialization is provided with the template correlation parameter, comprise the origin of template, the resolution of source sample and the resolution of sample to be tested, according to the resolution of source sample and the resolution of sample to be tested, the dynamic calculation sampling window, and take a sample from sample to be tested according to the sampling window that calculates;
S2. with the sampled picture binaryzation,, then the sampled picture after template binary image and the binaryzation is carried out and operation if template resolution is identical with sampling window resolution, remove the not point in template, obtain result images, otherwise, directly with the image as a result of of the image after the binaryzation;
S3. according to the relative value vector method, the space distribution vector of the result images that calculation procedure S2 obtains;
S4. adopt Euclidean distance method to carry out images match, ignoring result images space distribution vector during coupling is the value at 0 place at described template binary image vector correspondence position.
When sample to be tested resolution and source sample identical, sampling window is identical with template size to get final product, when sample to be tested resolution and source sample not simultaneously, then need dynamically to adjust the sampling window size according to sample to be tested resolution, so just can obtain best matching effect, particularly, as shown in Figure 2, step S1 further comprises:
S1.1 is if sample to be tested resolution is identical with the source sample resolution, and then sampling window is identical with template size, and execution in step S1.4, otherwise, according to the resolution calculating resolution variation scale factor of source sample and sample to be tested, and execution in step S1.2.Particularly, divided by the source sample resolution, obtain the change resolution scale factor with sample to be tested resolution.
S1.2 particularly, multiply by template resolution with this change resolution scale factor according to this change resolution scale factor calculation sampling window, obtains sampling window resolution.
S1.3 detects the variation of sample to be tested resolution in real time, if change, then returns step S1.1, otherwise, execution in step S1.4;
S1.4 takes a sample from sample to be tested according to sampling window.
After the sampling window that utilization is calculated by the dynamic window method among the step S1 is taken a sample, need to calculate the space distribution vector of sampled picture, be that (length of fixed length ordered series of numbers is preferably 64 among the present invention for a fixed length ordered series of numbers of representative image space distribution, be that vector is 64 dimensions), be used for follow-up calculating.Because sampling window is variable, so if utilize absolute value vector method compute vector, even result images and template matches so, the vectorial matching result that obtains also can be that the two does not match, therefore, the present invention adopts the relative value vector method.Particularly, as shown in Figure 3, step S3 further comprises:
S3.1 carries out image segmentation to the result images that step S2 obtains;
It is 255 to be the number of white point that S3.2 calculates in each cut zone pixel value;
S3.3 calculates the number percent that the white point number accounts for the pixel value of corresponding divided areas, obtains a decimal between the 0-1, described number percent is calculated the space distribution vector of described result images as vector element.
For not making effect simultaneously again, the efficient that improves method of the present invention do not descend too much, the present invention has adopted the Euclidean distance method in step S4, promptly calculate two Euclidean distances between vector, this method is simply efficient, and the credible result degree improves along with the raising of vectorial dimension, but consider for efficient, generally get 64 dimensional vectors and get final product.
Method of the present invention can use same template to detect the video of different resolution, is 100*100 such as the source sample resolution, has therefrom extracted template.The resolution of sample to be tested A is identical with the source sample, be 100*100, the resolution of sample to be tested B is 200*200, can use the template of from the sample of source, extracting to detect sample A and sample B, and need not need to extract template again for sample B again, when the sample to be tested change resolution, also can carry out faster and more accurate images match.
Above embodiment only is used to illustrate the present invention; and be not limitation of the present invention; the those of ordinary skill in relevant technologies field; under the situation that does not break away from the spirit and scope of the present invention; can also make various variations and modification; therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.
Claims (6)
1. image matching method is characterized in that the method comprising the steps of:
S1. according to the resolution of source sample and the resolution of sample to be tested, the dynamic calculation sampling window, and take a sample from described sample to be tested according to the sampling window that calculates;
S2. with the sampled picture binaryzation,, then the sampled picture after template binary image and the binaryzation is carried out and operation if template resolution is identical with sampling window resolution, remove the not point in template, obtain result images, otherwise, directly with the image as a result of of the image after the binaryzation;
S3. according to the relative value vector method, the space distribution vector of the result images that calculation procedure S2 obtains;
S4. adopt Euclidean distance method to carry out images match.
2. image matching method as claimed in claim 1 is characterized in that step S1 further comprises:
S1.1 is if sample to be tested resolution is identical with the source sample resolution, and then sampling window is identical with template size, and carries out S1.4, otherwise, according to the resolution calculating resolution variation scale factor of source sample and sample to be tested, and execution in step S1.2;
S1.2 is according to described change resolution scale factor calculation sampling window;
S1.3 detects the variation of sample to be tested resolution in real time, if change, then returns step S1.1, otherwise, execution in step S1.4;
S1.4 takes a sample from sample to be tested according to sampling window.
3. image matching method as claimed in claim 2 is characterized in that, in step S1.1, obtains described change resolution scale factor with described sample to be tested resolution divided by described source sample resolution.
4. image matching method as claimed in claim 2 is characterized in that, in step S1.2, multiply by template resolution with described change resolution scale factor, obtains sampling window resolution.
5. image matching method as claimed in claim 1 is characterized in that step S3 further comprises:
S3.1 carries out image segmentation to the result images that step S2 obtains;
S3.2 calculates the number of white point in each cut zone;
S3.3 calculates the number percent that the white point number accounts for the pixel value of corresponding divided areas, described number percent is calculated the space distribution vector of described result images as vector element.
6. the image matching method described in claim 1 is characterized in that, in step S4, ignoring described result images space distribution vector during coupling is the value at 0 place at described template binary image vector correspondence position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110175096 CN102214307B (en) | 2011-06-27 | 2011-06-27 | Image matching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110175096 CN102214307B (en) | 2011-06-27 | 2011-06-27 | Image matching method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102214307A true CN102214307A (en) | 2011-10-12 |
| CN102214307B CN102214307B (en) | 2013-01-16 |
Family
ID=44745607
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110175096 Active CN102214307B (en) | 2011-06-27 | 2011-06-27 | Image matching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102214307B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103514664A (en) * | 2012-06-14 | 2014-01-15 | 索尼公司 | Parking charging achievement device, system and method |
| CN113624952A (en) * | 2021-10-13 | 2021-11-09 | 深圳市帝迈生物技术有限公司 | In-vitro diagnosis device, detection method thereof and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101383007A (en) * | 2008-09-28 | 2009-03-11 | 腾讯科技（深圳）有限公司 | Image processing method and system based on integration histogram |
| CN101556695A (en) * | 2009-05-15 | 2009-10-14 | 广东工业大学 | Image matching method |
-
2011
- 2011-06-27 CN CN 201110175096 patent/CN102214307B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101383007A (en) * | 2008-09-28 | 2009-03-11 | 腾讯科技（深圳）有限公司 | Image processing method and system based on integration histogram |
| CN101556695A (en) * | 2009-05-15 | 2009-10-14 | 广东工业大学 | Image matching method |
Non-Patent Citations (1)
| Title |
|---|
| 《计算机技术与发展》 20090430 刘忠艳等 一种高效的图像匹配算法 45-47 1-6 第19卷, 第4期 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103514664A (en) * | 2012-06-14 | 2014-01-15 | 索尼公司 | Parking charging achievement device, system and method |
| CN103514664B (en) * | 2012-06-14 | 2015-12-09 | 索尼公司 | The implement device of parking charge, system and method |
| CN113624952A (en) * | 2021-10-13 | 2021-11-09 | 深圳市帝迈生物技术有限公司 | In-vitro diagnosis device, detection method thereof and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102214307B (en) | 2013-01-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105701476A (en) | Machine vision-based automatic identification system and method for production line products | |
| Pal et al. | Video shot boundary detection: a review | |
| CN115082419A (en) | Blow-molded luggage production defect detection method | |
| US20170061232A1 (en) | Describing objects using edge-pixel-feature descriptors | |
| CN102945378B (en) | Method for detecting potential target regions of remote sensing image on basis of monitoring method | |
| CN102426647A (en) | Station identification method and device | |
| CN103473551A (en) | Station logo recognition method and system based on SIFT operators | |
| CN102298781A (en) | Motion shadow detection method based on color and gradient characteristics | |
| CN103729631A (en) | Vision-based connector surface feature automatically-identifying method | |
| Tsintotas et al. | DOSeqSLAM: Dynamic on-line sequence based loop closure detection algorithm for SLAM | |
| Tiwari et al. | A survey on shadow detection and removal in images and video sequences | |
| CN110390682B (en) | Template self-adaptive image segmentation method, system and readable storage medium | |
| CN102378041B (en) | A kind of picture quality detection method and device | |
| CN105825201A (en) | Moving object tracking method in video monitoring | |
| CN102855473B (en) | A kind of image multi-target detection method based on similarity measurement | |
| CN102214307B (en) | Image matching method | |
| CN106844394B (en) | Video retrieval method based on pedestrian clothes and shirt color discrimination | |
| CN110363730A (en) | A kind of adaptive superpixel segmentation method based on color histogram peak value | |
| Iqbal et al. | Automated Meter Reading Detection Using Inception with Single Shot Multi-Box Detector. | |
| CN104504692A (en) | Method for extracting obvious object in image on basis of region contrast | |
| Piérard et al. | A probabilistic pixel-based approach to detect humans in video streams | |
| Wang et al. | Image segmentation based on support vector machine | |
| Hwang et al. | Enabling product recognition and tracking based on text detection for mobile augmented reality | |
| CN102184418A (en) | Triangle-area-representation-histogram-based image registration method | |
| Chuang et al. | Moving object segmentation and tracking using active contour and color classification models |
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 | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 100876 Beijing city Haidian District Xitucheng Road No. 10 Co-patentee after: BEIJING LANTUOPU TECHNOLOGY Co.,Ltd. Patentee after: Beijing University of Posts and Telecommunications Address before: 100876 Beijing city Haidian District Xitucheng Road No. 10 Co-patentee before: BEIJING BLUETOP TECHNOLOGY Co.,Ltd. Patentee before: Beijing University of Posts and Telecommunications |
|
| CP01 | Change in the name or title of a patent holder |