CN102779277A - Main vein extracting method based on image processing - Google Patents
Main vein extracting method based on image processing Download PDFInfo
- Publication number
- CN102779277A CN102779277A CN2012101896192A CN201210189619A CN102779277A CN 102779277 A CN102779277 A CN 102779277A CN 2012101896192 A CN2012101896192 A CN 2012101896192A CN 201210189619 A CN201210189619 A CN 201210189619A CN 102779277 A CN102779277 A CN 102779277A
- Authority
- CN
- China
- Prior art keywords
- texture
- main
- image
- main texture
- histogram
- 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
Images
Abstract
The invention discloses a main vein extracting method based on image processing. The method comprises the steps of: pre-processing an original graph, and converting the original graph into an HIS (Hue-Intensity-Saturation) color space for expression; performing histogram analysis, dividing the image through uniform grids with different dimensions, and counting the color histograms of vein blocks under various dimensions; extracting a main vein mask graph, comprehensively analyzing the results of the histograms to generate the main vein mask graph so as to determine the main vein; and synthesizing a vein effect and synthesizing the necessary large vein. The main vein extracting method disclosed by the invention analyzes the image and extracts the main components suitable for synthesis in the image, namely the main vein, so as to conveniently generate a high-quality synthesis effect.
Description
Technical field
The present invention relates to digital home technical field, be specifically related to a kind of based on the main texture method for distilling in the Flame Image Process.
Background technology
Texture is a kind of ubiquitous visual phenomenon, can be used to strengthen the sense of reality of image, have a wide range of applications in fields such as computer graphicss, as according to the synthetic large texture of sample texture, vector is visual and GIS-Geographic Information System etc.It is generally acknowledged that texture is to be made up of some repeated feature units, but these unit has certain randomness on COLOR COMPOSITION THROUGH DISTRIBUTION and size.Because texture is gathered mostly, tends to have some impurity contents, can have a strong impact on the efficient that texture uses from spontaneous phenomenon.For this reason, need from image pattern, will be suitable for the extracting section processing that texture calculates, reject impurity, and promptly carry out texture and extract.
Texture synthetic and to analyze be one of graph image hot research fields in recent years; Wherein the texture synthetic technology based on sample can generate visually similar large texture according to less sample texture; To reuse illumination information effectively, improve and draw efficient, be widely used.
It is very important extracting the main texture in the image to generating high-quality texture.The extraction of main texture mostly is in early days by the artificial completion of user, promptly in sample, chooses the image that meets textural characteristics by man-machine interactively ground, or in image, cuts out the part that is suitable for synthesizing.1999, the Markov probability model be introduced in texture synthetic in, obtained significant progress based on the texture synthetic technology of sample.2009, Lu etc. proposed a kind of based on the popular main texture detection method of diffusion length, can realize the automatic extraction of main texture.Some traditional image analytical approachs also can be used for carrying out the extraction of main texture like image partition method.
In early days carried out the image that mutual in sample, choosing meet textural characteristics or in image, cut out the part that is suitable for synthesizing by the method for manual work, though such operation is effective, extremely inconvenience wastes time and energy.Based on the texture of sample synthetic in, introduce the Markov model, then the color of arbitrary position is by the situation decision of its local finite neighborhood in the texture, just can progressively expand the generation with the completion target texture according to the neighborhood situation.The shortcoming of this method is if sample texture does not meet the characteristic of Markov probability model, will be difficult to generate high-quality texture according to this model.Therefore, the quality of sample becomes a key factor of the synthetic quality of restriction texture.
The method of Lu be follow according to main line unit in the image more tight with the more main line of the distance unit between the main line unit with the distance between the abnormity point, the therefore popular main texture that comes in the detection image of diffusion length capable of using.Through texture block all under a certain size in the image is analyzed, set up diffusion length stream shape, and find low dimension part as main texture based on the stream conformal analysis.This method computation process is complicated, and is very consuming time, and for only the have an appointment image of 10000 pixels of a width of cloth, this method is only set up stream shape needs 15 minutes with just analyzing.
The traditional image dividing method can come getting the different piece division in the image, can extract main texture thus.But such processing tends to the abnormity point that some are less to be covered in the main texture, and main texture also tends to be cut into several independent parts, has a strong impact on the quality that main texture extracts.
Summary of the invention
The objective of the invention is the defective in the conventional images processing,, avoid having influence on the quality that the style of writing extracts through extracting based on the main texture in the Flame Image Process.
It is a kind of based on the main texture method for distilling in the Flame Image Process that the present invention provides, and comprising:
Former figure is carried out pre-service, original image is transformed into the HIS color space representes;
Carry out histogram analysis, following that uniform grid is divided image with different size, add up the color histogram of texture block under each size;
Extract main texture mask figure, each histogrammic result of analysis-by-synthesis generates the mask figure of main texture, and then confirms main texture;
Synthetic texture result, synthetic needed large texture.
Saidly extract main texture mask figure and comprise: adopt the histogram on 32 rank to extract the main texture mask figure of image.
Said synthetic texture result comprises: according to the main texture mask figure that tries to achieve, block-by-block calculates former figure, tries to achieve main texture image.
Above technology can find out, algorithm of the present invention is through analyzing image, extract wherein to be suitable for the major part of synthesizing, and promptly main texture, thus can generate high-quality synthetic result easily.
Description of drawings
In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art; To do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art below; Obviously, the accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills; Under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.
Fig. 1 be in the embodiment of the invention based on the main texture method for distilling process flow diagram in the Flame Image Process;
Fig. 2 be in the embodiment of the invention not same order statistics with histogram this than synoptic diagram.
Embodiment
To combine the accompanying drawing in the embodiment of the invention below, the technical scheme in the embodiment of the invention is carried out clear, intactly description, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills are not making all other embodiment that obtained under the creative work prerequisite, all belong to the scope of the present invention's protection.
Often exist the texture cell that some repeat to occur in the main texture of image.Because main texture is common in occupation of bigger area in image, so the number of times that its texture block repeats to occur is also more than the impurity part.Thus, the color of texture block has higher value usually in the main texture in color histogram, and its distribution is also comparatively concentrated.If texture block is generated color histogram with its average color as representative, be positioned near the peak value the pairing texture block of color so and belong to main texture higher probability is arranged.
Fig. 1 show the present invention in implementing based on the main texture method for distilling process flow diagram in the Flame Image Process, comprise the steps:
S101: former figure is carried out pre-service, original image is transformed into the HIS color space representes;
S012: carry out histogram analysis, following that uniform grid is divided image with different size, add up the color histogram of texture block under each size;
S103: extract main texture mask figure, each histogrammic result of analysis-by-synthesis generates the mask figure of main texture, and then confirms main texture;
S104: synthetic texture result, synthetic needed large texture.
Wherein: HIS color space and component are selected
The difference of main texture and impurity can be embodied in many aspects in the image, might not only be embodied on the luminance difference, and the most frequently used histogram is the grey level histogram that is generated by brightness of image.For making main texture extract the visually-perceptible that more meets human eye, this paper adopts the widely used HIS color model in vision field that image is analyzed.
In Matlab, pass through transition matrix T=[1/sqrt (3) 1/sqrt (a 3) 1/sqrt (3);-sqrt (2)/6-sqrt (2)/6-sqrt (2)/6; 1/sqrt (2)-1/sqrt (2) 0], image is transformed into the HIS space by rgb space.
We find through a large amount of experimental analyses, only adopt H component and I component just to be enough to accomplish the extraction of main texture.The main texture that they help a part of image respectively extracts, and is unfavorable for extracting the image of another part but all images can carry out main efficiently texture with one of them component.
Generally speaking, if main texture has higher discrimination with the impurity part on a certain component, it all reflects this difference under the grid dividing situation of different size so.Correspond in the histogram, the different corresponding histogrammic distribution situations of size of dividing can reach unanimity.On the contrary, if main texture is higher with impurity similarity on a certain component, do not have tangible difference, this component is inappropriate for the texture of deciding and extracts so.Correspond in the histogram, this component distribution situation under different grid dividing situation also can have than big difference.Therefore, this patent calculate between each histogram of same component mean difference with, choose the foundation that difference and less component extract as main texture.
Multiple dimensioned grid dividing and histogram analysis
Given piece image I divides it with the uniform grid of M * N, generates the identical texture block of a series of sizes.The average pixel value of each texture block that calculating is marked off is as the representative color of this texture block, and adds up the gray level histogram Hist of these texture block representative colors
M * N(c).Known the color that belongs to the texture block of main texture has higher frequency in the histogram that generates, and should be in the part that has high value in the histogram and comparatively concentrated.Therefore can choose in the histogram near the part the peak value as the alternative area (C of main texture
Main)
M * N(C
Main)
M * NIn color need satisfy following 3 conditions:
1) histogrammic peak value is in this set, i.e. c
Max∈ (C
Main)
M * N, Hist
M * N(c
Max)=max (Hist
M * N(c), c wherein
MaxBe the highest color of frequency value in the histogram;
2) this is in non-color that is somebody's turn to do in the set mutually, and all colors in this set are at histogram middle distance c
MaxNearer, promptly the color in this set is in a concentrated area, and more balancedly is distributed in the peak value both sides;
3) α that the sum of pixel in image of color is greater than all number of pixels in the image in this set doubly, α is generally about 0.7.
After the color of having confirmed main texture alternative area, only need these colors are corresponded in the corresponding texture block of original image, can obtain the corresponding mask figure of main texture alternative area.
After the situation of having investigated multiple division, count which pixel and in most of the cases belonged to main texture alternative area, then just with the main texture of these pixel regions as extraction.Algorithm picks of the present invention 70% is as evaluation criterion, if promptly with 6 kinds of different sizes image is divided, the pixel that in 4 or above division, belongs to main texture alternative area so simultaneously can be finally as main texture.
Multistage histogram analysis and main texture mask figure extract
In statistics with histogram, the color-values of some impure points in the image also possibly have the characteristics of comparatively concentrating, but they generally can form comparatively sharp-pointed part.As shown in Figure 2, more sharp-pointed situation is arranged in the histogram on its 256 rank, in order to eliminate the influence of these impure points, can when histogram analysis, adopt the form of multiresolution.Promptly in the histogram of lower-order, analyze earlier, to eliminate the sharp-pointed information that indivedual impure point is caused.Basic confirm main texture region after, further in the histogram of higher-order, this zone is analyzed again, with the zone at the pixel value place of finding out main texture more accurately.Find through experiment, generally select the 2 kinds of different histogram resolution in 32 rank and 256 rank just can extract main texture effectively basically, can eliminate the interference of impurity information, can accurately locate main texture again.
Shown in Figure 2 be one not the same order statistics with histogram this, wherein be respectively the histogram on original image and 16,32,64,128 and 256 rank from left to right.From figure, observe visiblely, more sharp-pointed information is arranged in the histogram of high-order, this is brought by the impurity pixel in the image often; And exponent number when low some key character of image might lose, for example in 16 rank histograms, the left part of histogram peak has disappeared, and has produced bigger error.Find that through experiment 32 rank are these comparatively ideal selections, can eliminate sharp-pointed information, are unlikely to produce bigger error again.So the histogram that adopts 32 rank extracts the main texture mask figure of image.
Synthetic large texture
According to the main texture mask figure that tries to achieve, block-by-block calculates original image, tries to achieve main texture image.Main texture image is duplicated expansion, can obtain required large texture image.
The embodiment of the invention at first is transformed into the HIS color space with original image and representes; With the uniform grid of different size image is divided afterwards, added up the color histogram of texture block under each size.There is bigger probability to belong to main texture owing in each histogram, be in the texture block of concentrated area, therefore can generates the mask figure of main texture, and then confirm main texture through each histogrammic result of analysis-by-synthesis.
The present invention proposes a kind of simple and direct main texture method for distilling, can fast and effeciently extract the main texture in the image.This method mainly is based on the HIS color model image color is carried out histogrammic statistical presentation; Visually-perceptible with the reflection human eye; And generally in sample image, occupy the characteristics in most of zone according to main texture, can carry out the extraction of main texture based on easy statistical study.With present known the best way mutually this, this paper method can improve 4 one magnitude with speed, can on general microcomputer, handle a width of cloth sample image with the time that is less than 1s, and the main texture of its extraction is suitable with existing work, can generate high-quality large texture.
One of ordinary skill in the art will appreciate that all or part of step in the whole bag of tricks of the foregoing description is to instruct relevant hardware to accomplish through program; This program can be stored in the computer-readable recording medium; Storage medium can comprise: ROM (read-only memory) (ROM; Read Only Memory), RAS (RAM, RandomAccess Memory), disk or CD etc.
More than the main texture method for distilling based in the Flame Image Process that the embodiment of the invention provided has been carried out detailed introduction; Used concrete example among this paper principle of the present invention and embodiment are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that on embodiment and range of application, all can change, in sum, this description should not be construed as limitation of the present invention.
Claims (3)
1. one kind based on the main texture method for distilling in the Flame Image Process, it is characterized in that said method comprises:
Former figure is carried out pre-service, original image is transformed into the HIS color space representes;
Carry out histogram analysis, following that uniform grid is divided image with different size, add up the color histogram of texture block under each size;
Extract main texture mask figure, each histogrammic result of analysis-by-synthesis generates the mask figure of main texture, and then confirms main texture;
Synthetic texture result, synthetic needed large texture.
2. as claimed in claim 1ly it is characterized in that, saidly extract main texture mask figure and comprise based on the main texture method for distilling in the Flame Image Process:
Adopt the histogram on 32 rank to extract the main texture mask figure of image.
3. as claimed in claim 1ly it is characterized in that based on the main texture method for distilling in the Flame Image Process said synthetic texture result comprises:
According to the main texture mask figure that tries to achieve, block-by-block calculates former figure, tries to achieve main texture image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101896192A CN102779277A (en) | 2012-06-08 | 2012-06-08 | Main vein extracting method based on image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101896192A CN102779277A (en) | 2012-06-08 | 2012-06-08 | Main vein extracting method based on image processing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102779277A true CN102779277A (en) | 2012-11-14 |
Family
ID=47124187
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012101896192A Pending CN102779277A (en) | 2012-06-08 | 2012-06-08 | Main vein extracting method based on image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102779277A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103529939A (en) * | 2013-09-30 | 2014-01-22 | 北京航空航天大学 | Man-machine interaction method based on color, computer vision and grid division |
| CN109859257A (en) * | 2019-02-25 | 2019-06-07 | 北京工商大学 | A kind of skin image texture appraisal procedure and system based on grain direction |
| WO2023179465A1 (en) * | 2022-03-24 | 2023-09-28 | 张国流 | Image texture extraction method and device, and computer readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090185747A1 (en) * | 2008-01-18 | 2009-07-23 | Sharp Laboratories Of America, Inc. | Systems and methods for texture synthesis for video coding with side information |
| CN101556600A (en) * | 2009-05-18 | 2009-10-14 | 中山大学 | Method for retrieving images in DCT domain |
| CN101937567A (en) * | 2010-09-28 | 2011-01-05 | 中国科学院软件研究所 | Method for extracting main texture simply and quickly |
-
2012
- 2012-06-08 CN CN2012101896192A patent/CN102779277A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090185747A1 (en) * | 2008-01-18 | 2009-07-23 | Sharp Laboratories Of America, Inc. | Systems and methods for texture synthesis for video coding with side information |
| CN101556600A (en) * | 2009-05-18 | 2009-10-14 | 中山大学 | Method for retrieving images in DCT domain |
| CN101937567A (en) * | 2010-09-28 | 2011-01-05 | 中国科学院软件研究所 | Method for extracting main texture simply and quickly |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103529939A (en) * | 2013-09-30 | 2014-01-22 | 北京航空航天大学 | Man-machine interaction method based on color, computer vision and grid division |
| CN103529939B (en) * | 2013-09-30 | 2016-08-17 | 北京航空航天大学 | A kind of man-machine interaction method based on color, computer vision and stress and strain model |
| CN109859257A (en) * | 2019-02-25 | 2019-06-07 | 北京工商大学 | A kind of skin image texture appraisal procedure and system based on grain direction |
| WO2023179465A1 (en) * | 2022-03-24 | 2023-09-28 | 张国流 | Image texture extraction method and device, and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102496023B (en) | Region of interest extraction method of pixel level | |
| CN110796038B (en) | Hyperspectral remote sensing image classification method combined with rapid region growing superpixel segmentation | |
| Machairas et al. | Waterpixels: Superpixels based on the watershed transformation | |
| JP2009157821A (en) | Range image generating device, environment recognition device, and program | |
| CN105404657A (en) | CEDD feature and PHOG feature based image retrieval method | |
| Kanjanawanishkul et al. | Novel fast color reduction algorithm for time-constrained applications | |
| US20110134128A1 (en) | Visualization and Representation of Data Clusters and Relations | |
| Yan et al. | Codebook guided feature-preserving for recognition-oriented image retargeting | |
| CN102779277A (en) | Main vein extracting method based on image processing | |
| Han et al. | An enhanced image binarization method incorporating with Monte-Carlo simulation | |
| CN103761736A (en) | Image segmentation method based on Bayes harmonious degree | |
| Umaselvi et al. | Color based urban and agricultural land classification by GLCM texture features | |
| CN103714517A (en) | Video rain removing method | |
| Senturk et al. | Seam carving based image retargeting: A survey | |
| CN110310263B (en) | SAR image residential area detection method based on significance analysis and background prior | |
| US8644607B1 (en) | Method and apparatus for image processing | |
| CN101853494B (en) | Color image segmentation method based on coring fuzzy Fisher criterion clustering | |
| CN104331883A (en) | Image boundary extraction method based on non-symmetry and anti-packing model | |
| CN105205485B (en) | Large scale image partitioning algorithm based on maximum variance algorithm between multiclass class | |
| Malladi et al. | Superpixels using morphology for rock image segmentation | |
| Shahiduzzaman et al. | Improved spatial pyramid matching for image classification | |
| CN101937567B (en) | Method for extracting main texture simply and quickly | |
| CN106485686A (en) | One kind is based on gravitational spectral clustering image segmentation algorithm | |
| Quweider et al. | Grey level reduction for segmentation, threshholding and binarisation of images based on optimal partitioning on an interval | |
| Partio et al. | An ordinal co-occurrence matrix framework for texture retrieval |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20121114 |