CN105654449A - Video image discrete division method - Google Patents
Video image discrete division method Download PDFInfo
- Publication number
- CN105654449A CN105654449A CN201410623319.XA CN201410623319A CN105654449A CN 105654449 A CN105654449 A CN 105654449A CN 201410623319 A CN201410623319 A CN 201410623319A CN 105654449 A CN105654449 A CN 105654449A
- Authority
- CN
- China
- Prior art keywords
- image
- subgraph
- block
- sub
- segmentation
- 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
Landscapes
- Image Analysis (AREA)
Abstract
A video image discrete division method belongs to the technical field of video division methods, and especially relates to a video image discrete division method. The invention provides a video image discrete division method which is quick in image division. The video image discrete division method comprises the following steps: (1) dividing a whole image into a plurality of sub image blocks on the basis of window size; (2) extracting the feature vectors of the sub images, carrying out discrete stationary wavelet transform on the sub images, and working out the average energy of each sub image in each wavelet domain as the feature value of the sub image; and (3) pre-dividing the whole image in units of sub image blocks: extracting the feature values of the sub image blocks, and pre-dividing the image by using a fuzzy clustering method.
Description
Technical field
The invention belongs to video division methods technical field, particularly relate to a kind of discrete division methods of video image.
Background technology
Texture image presents scrambling in regional area, and it is regular to show certain on the whole. The arrangement of texture primitive is probably random, it is also possible to interdepend each other, and this dependency is probably structure, it is also possible to arrange by certain probability distribution, it is also possible to certain functional form. Image texture can describe with many language qualitatively, such as coarse, fine, smooth, directivity and systematicness, granularity etc. But it is not a nothing the matter that these semantemes change into mathematical model. About the texture analysis having Hust coefficient, gray level co-occurrence matrixes and texture energy that the method for Texture Segmentation is conventional. Substantially retrieve with statistics and filtering method for main texture feature extraction at present.
Summary of the invention
The present invention is aiming at the problems referred to above, it is provided that the discrete division methods of video image that a kind of image segmentation speed is fast.
For achieving the above object, the present invention comprises the following steps.
1) whole figure is divided in units of window size some subgraph blocks.
2) extract the characteristic vector of subgraph, after subgraph is carried out Stationary Wavelet Transform, each subgraph is obtained by each wavelet field this subgraph average energy as this block feature amount.
3) in units of subimage block, whole figure is carried out pre-segmentation; After block under subgraph is extracted characteristic quantity, by adopting fuzzy clustering method, image is carried out pre-segmentation.
4) determine the sub-block of coarse segmentation edge, each pixel in edge sub-block is carried out feature extraction by moving window, the part at edge is finely divided and cuts; Before classifying, the eigenvalue of each character vector is normalized according to the following formula.
As a kind of preferred version, present invention additionally comprises: image is carried out Stationary Wavelet Transform two grades decomposition, obtain counting identical wavelet coefficient with original image in each subband.
Original image is divided into several subimages, in units of subgraph block, in each little wavestrip of small echo, seeks its average energy according to the formula of laws energy, and as characteristic vector, adopt FuzzycMeans Clustering method that image is carried out coarse segmentation.
As another kind of preferred version, present invention additionally comprises: find the border sub-block of coarse segmentation, in these sub-blocks, in units of pixel, carry out feature extraction, again edge subimage block is carried out cluster point segmentation.
Beneficial effect of the present invention.
The present invention adopt steady wavelet transform and FuzzycMeans Clustering and improve unity and coherence in writing image is split by dividing method, by actual effect and data relatively, utilizing algorithm provided by the invention to improve the image segmentation speed of dual texture, segmentation result is closer to practical situation.
Detailed description of the invention
The present invention comprises the following steps.
1) whole figure is divided in units of window size some subgraph blocks.
2) extract the characteristic vector of subgraph, after subgraph is carried out Stationary Wavelet Transform, each subgraph is obtained by each wavelet field this subgraph average energy as this block feature amount.
3) in units of subimage block, whole figure is carried out pre-segmentation; After block under subgraph is extracted characteristic quantity, by adopting fuzzy clustering method, image is carried out pre-segmentation.
4) determine the sub-block of coarse segmentation edge, each pixel in edge sub-block is carried out feature extraction by moving window, the part at edge is finely divided and cuts; Before classifying, the eigenvalue of each character vector is normalized according to the following formula.
Present invention additionally comprises: image is carried out Stationary Wavelet Transform two grades decomposition, obtain counting identical wavelet coefficient with original image in each subband.
Original image is divided into several subimages, in units of subgraph block, in each little wavestrip of small echo, seeks its average energy according to the formula of laws energy, and as characteristic vector, adopt FuzzycMeans Clustering method that image is carried out coarse segmentation.
Present invention additionally comprises: find the border sub-block of coarse segmentation, in these sub-blocks, in units of pixel, carry out feature extraction, again edge subimage block is carried out cluster point segmentation.
The present invention is split with the figure in texture searching for object, and small echo adopts daubechies small echo series db3. Image is sized to 256 �� 256. Window size is 16 �� 16. Before segmentation and after segmentation, image is not done any process.
When adopting the secondary splitting method that normal segmentation method and the present invention propose, the iterations extracting characteristic vector in any one wavelet field compares.
Conversion process exists the down-sampled problem to sampled point so that wavelet transformation loses translation invariant characteristic. The translation invariance of small echo is very important for processing the application of statistical signal, and therefore, this is unfavorable for that wavelet transformation extracts edge, textural characteristics application in image is split. Present invention employs steady wavelet transform for this, compared with classical Discrete Orthogonal Wavelet Transform, steady wavelet transform be mainly characterized by redundancy and translation invariance, continuous wavelet transform can be provided a more approximate estimation.
Steady wavelet transform is different according to occasion its call different from purpose of its application, such as translation invariant wavelet, stationary wavelet etc. In Stationary Wavelet Transform process, the problem being absent from down-sampling, the coefficient length that therefore stationary wavelet decomposes every time is identical with the length of primary signal, it is possible to better spatially corresponding with primary signal. Being more beneficial for processing the signal with statistical law, therefore the present invention has selected Stationary Wavelet Transform as the instrument of Study Of Segmentation Of Textured Images. For image procossing commonly used be two-dimensional discrete wavelet conversion, two-dimensional wavelet transformation is the popularization of one-dimensional wavelet transform.
Above content is the further description present invention made in conjunction with concrete preferred implementation; it cannot be assumed that specific embodiment of the invention is confined to these explanations; for general technical staff of the technical field of the invention; without departing from the inventive concept of the premise; some simple deduction or replace can also be made, all should be considered as belonging to the protection domain that the submitted claims of the present invention are determined.
Claims (3)
1.A kind of discrete division methods of video image, it is characterised in that comprise the following steps:
1) whole figure is divided in units of window size some subgraph blocks;
2) extract the characteristic vector of subgraph, after subgraph is carried out Stationary Wavelet Transform, each subgraph is obtained by each wavelet field this subgraph average energy as this block feature amount;
3) in units of subimage block, whole figure is carried out pre-segmentation; After block under subgraph is extracted characteristic quantity, by adopting fuzzy clustering method, image is carried out pre-segmentation;
4) determine the sub-block of coarse segmentation edge, each pixel in edge sub-block is carried out feature extraction by moving window, the part at edge is finely divided and cuts; Before classifying, the eigenvalue of each character vector is normalized according to the following formula.
2.According to claim 1, a kind of discrete division methods of video image, it is characterised in that also include: image carries out Stationary Wavelet Transform two grades decomposition, obtain counting identical wavelet coefficient with original image in each subband;
Original image is divided into several subimages, in units of subgraph block, in each little wavestrip of small echo, seeks its average energy according to the formula of laws energy, and as characteristic vector, adopt FuzzycMeans Clustering method that image is carried out coarse segmentation.
3.According to claim 2, a kind of discrete division methods of video image, it is characterised in that also include: find the border sub-block of coarse segmentation, in these sub-blocks, carry out feature extraction in units of pixel, carry out cluster point segmentation again to edge subimage block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410623319.XA CN105654449A (en) | 2014-11-09 | 2014-11-09 | Video image discrete division method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410623319.XA CN105654449A (en) | 2014-11-09 | 2014-11-09 | Video image discrete division method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105654449A true CN105654449A (en) | 2016-06-08 |
Family
ID=56483364
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410623319.XA Pending CN105654449A (en) | 2014-11-09 | 2014-11-09 | Video image discrete division method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105654449A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127813A (en) * | 2016-07-04 | 2016-11-16 | 石家庄铁道大学 | The monitor video motion segments dividing method of view-based access control model energy sensing |
| CN116012607A (en) * | 2022-01-27 | 2023-04-25 | 华南理工大学 | Image weak texture feature extraction method and device, equipment and storage medium |
-
2014
- 2014-11-09 CN CN201410623319.XA patent/CN105654449A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127813A (en) * | 2016-07-04 | 2016-11-16 | 石家庄铁道大学 | The monitor video motion segments dividing method of view-based access control model energy sensing |
| CN106127813B (en) * | 2016-07-04 | 2018-04-10 | 石家庄铁道大学 | The monitor video motion segments dividing method of view-based access control model energy sensing |
| CN116012607A (en) * | 2022-01-27 | 2023-04-25 | 华南理工大学 | Image weak texture feature extraction method and device, equipment and storage medium |
| CN116012607B (en) * | 2022-01-27 | 2023-09-01 | 华南理工大学 | Image weak texture feature extraction method and device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101859320A (en) | Massive image retrieval method based on multi-characteristic signature | |
| CN104850859A (en) | Multi-scale analysis based image feature bag constructing method | |
| Liu et al. | Study on rotation-invariant texture feature extraction for tire pattern retrieval | |
| Ma et al. | Log-can: local-global class-aware network for semantic segmentation of remote sensing images | |
| Lasfar et al. | Content-based retrieval in fractal coded image databases | |
| CN105654449A (en) | Video image discrete division method | |
| CN106886613A (en) | A kind of Text Clustering Method of parallelization | |
| Pawar et al. | Region based image classification using watershed transform techniques | |
| Ansari et al. | Textural classification based on wavelet, curvelet and contourlet features | |
| CN103077381A (en) | Monocular dynamic hand gesture recognition method on basis of fractional Fourier transformation | |
| CN108805183A (en) | A kind of image classification method of fusion partial polymerization descriptor and local uniform enconding | |
| CN104376300A (en) | Identification method used for intelligent matching of incomplete Chinese characters on basis of grid characteristics | |
| Kang et al. | Weak constraint leaf image recognition based on convolutional neural network | |
| Kothyari et al. | Content based image retrieval using statistical feature and shape extraction | |
| Xu et al. | Fast Feature Extraction Based on Multi-feature Classification for Color Image. | |
| CN110019904A (en) | A kind of image search method based on big data | |
| CN106778535B (en) | Iris feature extraction and matching method based on wavelet packet decomposition | |
| Rajakumar et al. | MRI image retrieval using wavelet with Mahalanobis distance measurement | |
| Latha et al. | Content based color image retrieval via wavelet transforms | |
| Wang et al. | Digital image retrieval technology based on multi-feature | |
| Ma et al. | Texture image segmentation on improved watershed and multiway spectral clustering | |
| Chen et al. | Quadtree classified vector quantization based image retrieval scheme | |
| Wang et al. | Image watermarking algorithm based on grey relational analysis and singular value decomposition in wavelet domain | |
| Yan et al. | An improved texture feature extraction method based on radon transform | |
| Chaker et al. | Exploiting disparity information for stereo image retrieval |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160608 |