CN102567969A - Color image edge detection method - Google Patents
Color image edge detection method Download PDFInfo
- Publication number
- CN102567969A CN102567969A CN2011104481191A CN201110448119A CN102567969A CN 102567969 A CN102567969 A CN 102567969A CN 2011104481191 A CN2011104481191 A CN 2011104481191A CN 201110448119 A CN201110448119 A CN 201110448119A CN 102567969 A CN102567969 A CN 102567969A
- Authority
- CN
- China
- Prior art keywords
- image
- color
- antagonism
- edge
- marginal information
- 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
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a color image edge detection method which comprises the following specific steps: extracting each color channel image, generating color antagonism images, calculating an edge information distribution image, and refining the color image edge. According to the method, a color image is discomposed into a red channel image, a green channel image, a blue channel image and a yellow channel image, red-green antagonism images and blue-yellow antagonism images are calculated respectively, edge information distribution images are obtained on the antagonism images through calculating, and the strength of the color edge and the brightness edge are adjusted conveniently through two groups of antagonism weighting factors introduced in the process of adjusting, calculating and generating the color antagonism images, thus the color image edge can be detected flexibly and effectively. The color image edge detection method has the advantages that the brightness or color edge information of the nature color image can be effectively detected through simple parameter selection, and in addition, the brightness and color edge information of the nature color image can be selectively extracted.
Description
Technical field
The invention belongs to technical field of computer vision, particularly the rim detection of coloured image.
Background technology
Coloured image provides than the gray level image information of horn of plenty more, so Color Image Processing is just receiving increasing concern.In the natural image; Understanding all has vital role for scene for color and luminance edges information; In order from complicated natural image, effectively to extract marginal information; Two types of color image edge detection methods of main at present existence: class methods are each passage use gray scale edge detection methods at coloured image, and this method can not effectively detect color edges; Other methods at first are transformed into the shades of colour space with coloured image, and calculate color and luminance edges respectively, remerge colouring information and monochrome information is extracted edge of image.Relatively typical method has the method for propositions in 2004 such as Martin, referring to document: D.R.Martin, and C.C.Fowlkes and J.Malik.Learning to detect natural image boundaries using local brightness; Color; And texture cues.Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2004; 26 (5): 530-549; This method at first with image transitions to the CIELab space, respectively in L, a, b passage compute gradient information, thereby can effectively detect color and luminance edges.But the major defect of this method is must extract brightness and colouring information respectively in order to extract the marginal information of coloured image, and then merges through the training weighting, obtains edge image, its calculation of complex, very flexible.
Summary of the invention
The objective of the invention is to have proposed a kind of color image edge detection method in order to solve the defective that existing above-mentioned color image edge detection method exists.
To achieve these goals, technical scheme of the present invention is: a kind of color image edge detection method comprises the steps:
S1. extract each Color Channel image: the coloured image of input is decomposed into redness, green and blue three Color Channel images; Again with the mean value of red, green channel image as the yellow channels image;
S2. generate color antagonism image: redness and green channel image that step S1 is obtained are multiplied by predefined two antagonism weight factors respectively, and addition obtains red green antagonism image; Same, through predefined two other antagonism weight factor, utilize blueness and yellow channels image calculation to obtain blue yellow antagonism image;
S3. edge calculation information distribution image: the red green antagonism image and the blue yellow antagonism image that obtain of calculation procedure S2 respectively; Obtain red green marginal information distributed image and blue yellow marginal information distributed image; Get the maximal value of red green and blue yellow marginal information distributed image correspondence position, obtain the marginal information distributed image;
S4. edge thinning is handled: with the edge thinning method marginal information distributed image that step S3 obtains is handled, obtained final edge image.
Step S3 is described, and to obtain the concrete computation process of red green marginal information distributed image and blue yellow marginal information distributed image following: make up the gradient computing template under a plurality of directions; The red green antagonism image that respectively step S2 is obtained carries out Filtering Processing; And each gray values of pixel points in the filtering output image taken absolute value, obtain the downward gradient information distribution plan of counterparty; Choose the marginal information intensity of the maximal value of correspondence position in the gradient information distribution plan under each direction again, obtain red green marginal information distributed image as this position; According to same computing method the yellow antagonism image of indigo plant is handled and to be obtained blue yellow marginal information distributed image.
As a preferred embodiment, step S1 also comprises and utilizes smoothing filter four channel image to be carried out the process of smoothing processing.
The span of the scale parameter value of above-mentioned smoothing filter is 1～5.
A plurality of directions of the gradient computing described in the above-mentioned steps S3 are specially: in 180 degree scopes, wait radian to distribute, the direction number is 8～18.
Edge thinning method described in the above-mentioned steps S4 is specially non-maximum value inhibition method.
Beneficial effect of the present invention: method of the present invention at first decomposes redness, green, blueness and yellow four channel image with coloured image; Calculate red green, blue yellow antagonism image then respectively; And on the antagonism image, calculate the marginal information distributed image; Through two groups of antagonism weight factors introducing among the regulating step S2, can regulate the intensity of color edges and luminance edges in the image easily, realize rim detection flexibly and effectively.When calculating red green antagonism image, when two antagonism weight factor jack per lines, this method is beneficial to the sensed luminance edge, and when two antagonism weight factors equated, only detection had the edge of difference in brightness especially; On the contrary, when two antagonism weight factor contrary signs, this method is beneficial to the detection color edges, especially when two antagonism weight factor opposite in signs, during equal and opposite in direction, only detects the edge that color distinction is arranged.When calculating blue yellow antagonism image also is like this.In addition, among the step S2,, can also optionally strengthen red green marginal information or blue yellow marginal information for red calculating green and blue yellow antagonism image is provided with two groups of different antagonism weight factors respectively.Detection method of the present invention has through simple parameter to be selected, and effectively detects brightness in the natural color image, color edges information, also can extract brightness or color edges information selectively simultaneously.Method of the present invention has the characteristics such as simple, effective flexibly of calculating.
Description of drawings
Fig. 1 is the schematic flow sheet of a kind of color image edge detection method of the present invention.
Fig. 2 is an antagonism weight factor selection and skirt response relation synoptic diagram.
Fig. 3 be adopt among the embodiment the inventive method to colored natural image carry out actual detected outline map and with the comparison diagram group of standard edge figure, luminance edges figure and color edges figure.
Embodiment
Below in conjunction with accompanying drawing and concrete embodiment the present invention is done further elaboration.
The human visual system utilizes color antagonism principle that color and monochrome information are had the ability of processing simultaneously.The elementary cell that the receptive field of optic nerve unit is handled as the vision system colouring information, red green, the blue yellow color antagonistic properties that when handling chromatic information, has has proposed method of the present invention based on this.
Specify through an embodiment below.
Download 101087 images and corresponding nominal contour testing result thereof from the image library website of present internationally recognized checking profile extraction algorithm effect; The image size is 481 * 321; Wherein 101087 images are 24 true color images, and stack obtained after the nominal contour testing result was delineated by a plurality of people are manual.The flow process of concrete detection method is as shown in Figure 1, and detailed process is following:
S1. extract each Color Channel image: at first the coloured image with input is decomposed into redness, green and blue three Color Channel images; The mean value that calculates red, green passage again is as the yellow channels image.With pixel (200; 200) be example, this pixel is respectively 99,82 and 30 at red, green, blue passage gray-scale value, and this position grey scale pixel value is the mean value of red green passage correspondence position grey scale pixel value in the yellow channels; Be that the value of pixel (200,200) in yellow channels is 90.5; Utilizing scale parameter is that Gauss's smoothing filter of 1.5 carries out smoothing processing respectively to four channel image.The gray-scale value of pixel after the smoothing processing (200,200) in red, green, blue and yellow passage is respectively 113.5720,97.6701,41.4333 and 105.6201.
Here each channel image being carried out The disposal of gentle filter respectively is an optional process; Purpose is to suppress effectively the noise of image; Can select suitable smoothing filter yardstick according to actual needs, in general, the span of scale parameter value is 1～5.
S2. generate color antagonism image: will be multiplied by predefined two antagonism weight factors respectively through the redness (R) after step S1 gained is level and smooth and green (G) channel image respectively, addition obtains red green antagonism image.
Here, the value of antagonism weight factor is not except that being that scope is not limit, and can set according to demand zero simultaneously.
Preestablishing redness and green channel image antagonism weight factor in the present embodiment is respectively: ω
1=1.0, ω
2=-0.5; Again through calculating I
Rg=ω
1R+ ω
2G obtains red green antagonism image (I
Rg); With pixel (200,200) is example, and this some gray-scale value in red green antagonism image is 1.0 * 113.5720+ (0.5) * 97.6701=64.7370.
Same process is through predefined two other antagonism weight factor ω
3, ω
4, blueness and yellow channels image calculation after utilization is level and smooth obtain blue yellow antagonism image.
Fig. 2 explains the influence of the selection of antagonism weight factor to skirt response.Expressing for convenient, is example with luminance edges and pure red green edge only, simultaneously fixing ω
1=1.Visible by figure, work as ω
1With ω
2During jack per line, the response of luminance edges is worked as ω especially greater than color edges
1=ω
2, the color edges response is 0; Work as ω
1With ω
2During contrary sign, the response of color edges is worked as greater than luminance edges especially | ω
1|=| ω
2|, the luminance edges response is 0.Based on this, can select the antagonism weight factor flexibly according to demand.
One group of antagonism weight factor that the red green antagonism image of above-mentioned generation is used with generate one group of antagonism weight factor that blue yellow antagonism image uses can be identical; Also can be different; In the present embodiment; For convenience of calculation be convenient to say something, used two groups of identical antagonism weight factors, promptly preestablish red and green channel image antagonism weight factor ω
1, ω
2Be respectively: ω
1=1.0, ω
2=-0.5; Preestablish blue and yellow channels image antagonism weight factor ω
3, ω
4Respectively with ω
1, ω
2Identical, promptly value is respectively: ω
3=ω
1=1.0, ω
4=ω
2=-0.5.In the process that obtains Fig. 3 c, Fig. 3 d, what use also is two groups of identical antagonism weight factors.
S3. edge calculation information distribution image: utilize the first order derivative of Gaussian function to make up the gradient computing template under 8 directions; And respectively the red green image that calculates through step S2 is carried out Filtering Processing;-5.0702 ,-2.7997,0.3482,2.6581,4.1599,5.0618,5.3494,5.4556 for example pixel (200,200) filtered gray-scale value under all directions is respectively:; Again each gray-scale value is taken absolute value and be respectively 5.0702,2.7997,0.3482,2.6581,4.1599,5.0618,5.3494,5.4556; To each pixel in the image, obtain 8 gradient information distribution plans under the direction according to the same manner; Choose the marginal information intensity of the maximal value of correspondence position in the gradient information distribution plan under each direction as this position again, obtain red green marginal information distributed image, wherein, the gray-scale value that pixel (200,200) is located is 5.4556; According to same computing method, on the yellow antagonism image of indigo plant, calculate blue yellow marginal information distributed image, wherein, the gray-scale value that pixel (200,200) is located is 2.3568; Get the maximal value of red green and blue yellow marginal information distributed image correspondence position, obtain the marginal information distributed image, wherein, the gray-scale value that pixel (200,200) is located is 5.4556.
Can find out, also can use other method edge calculation information distribution image here.
S4. edge thinning is handled: the marginal information distributed image to step S3 obtains is handled, and obtains single pixel wide outline map, can obtain final edge image.
Here specifically can adopt non-maximum value inhibition method.The edge intensity value computing scope is 0-1, and wherein, the gray-scale value that pixel (200,200) is located is 0.
Among the step S2, select different antagonism weight factors can control brightness or color edges, for example antagonism weight factor ω flexibly
1=1.0, ω
2=1.0 o'clock, a sensed luminance edge, final detected edge image is shown in Fig. 3 c; Antagonism weight factor ω
1=1.0, ω
2=-1.0 o'clock, only detect color edges, final detected edge image is shown in Fig. 3 d; Antagonism weight factor ω
1=1.0, ω
2=-0.5 o'clock, can detect brightness and color edges simultaneously, final detected edge image is shown in Fig. 3 e;
Test result is as shown in Figure 3; The black table indicating value is 1 among the figure, and white expression value is 0, wherein: the 3a. original image; 3b. standard edge image; 3c. the detected luminance edges image of the present invention, 3d. the present invention detects the color edges image, and 3e. the present invention detects edge brightness and color edges image simultaneously.From figure, can know and find out, when adopting the inventive method to extract Color Image Edge, can select sensed luminance edge or color edges flexibly, and while sensed luminance and color edges, it calculates simple, flexibly.
Those of ordinary skill in the art will appreciate that embodiment described here is in order to help reader understanding's principle of the present invention, should to be understood that protection scope of the present invention is not limited to such special statement and embodiment.Those of ordinary skill in the art can make various other various concrete distortion and combinations that do not break away from essence of the present invention according to these teachings disclosed by the invention, and these distortion and combination are still in protection scope of the present invention.
Claims (6)
1. a color image edge detection method is characterized in that, comprises the steps:
S1. extract each Color Channel image: the coloured image of input is decomposed into redness, green and blue three Color Channel images; Again with the mean value of red, green channel image as the yellow channels image;
S2. generate color antagonism image: redness and green channel image that step S1 is obtained are multiplied by predefined two antagonism weight factors respectively, and addition obtains red green antagonism image; Same, through predefined two other antagonism weight factor, utilize blueness and yellow channels image calculation to obtain blue yellow antagonism image;
S3. edge calculation information distribution image: the red green antagonism image and the blue yellow antagonism image that obtain of calculation procedure S2 respectively; Obtain red green marginal information distributed image and blue yellow marginal information distributed image; Get the maximal value of red green and blue yellow marginal information distributed image correspondence position, obtain the marginal information distributed image;
S4. edge thinning is handled: with the edge thinning method marginal information distributed image that step S3 obtains is handled, obtained final edge image.
2. color image edge detection method according to claim 1; It is characterized in that; Step S3 is described, and to obtain the concrete computation process of red green marginal information distributed image and blue yellow marginal information distributed image following: make up the gradient computing template under a plurality of directions; The red green antagonism image that respectively step S2 is obtained carries out Filtering Processing, and each gray values of pixel points in the filtering output image is taken absolute value, and obtains the downward gradient information distribution plan of counterparty; Choose the marginal information intensity of the maximal value of correspondence position in the gradient information distribution plan under each direction again, obtain red green marginal information distributed image as this position; According to same computing method the yellow antagonism image of indigo plant is handled and to be obtained blue yellow marginal information distributed image.
3. color image edge detection method according to claim 1 and 2 is characterized in that, step S1 also comprises and utilizes smoothing filter four channel image to be carried out the process of smoothing processing.
4. color image edge detection method according to claim 3 is characterized in that, the span of the scale parameter value of said smoothing filter is 1～5.
5. color image edge detection method according to claim 4 is characterized in that, a plurality of directions of the gradient computing described in the step S3 are specially: in 180 degree scopes, wait radian to distribute, the direction number is 8～18.
6. according to claim 4 or 5 described color image edge detection methods, it is characterized in that the edge thinning method described in the step S4 is specially non-maximum value inhibition method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110448119.1A CN102567969B (en) | 2011-12-28 | 2011-12-28 | Color image edge detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110448119.1A CN102567969B (en) | 2011-12-28 | 2011-12-28 | Color image edge detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102567969A true CN102567969A (en) | 2012-07-11 |
| CN102567969B CN102567969B (en) | 2014-06-18 |
Family
ID=46413320
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110448119.1A Active CN102567969B (en) | 2011-12-28 | 2011-12-28 | Color image edge detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102567969B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105139391A (en) * | 2015-08-17 | 2015-12-09 | 长安大学 | Edge detecting method for traffic image in fog-and-haze weather |
| CN105957067A (en) * | 2016-04-23 | 2016-09-21 | 北京工业大学 | Color image edge detection method based on color difference |
| CN106228547A (en) * | 2016-07-15 | 2016-12-14 | 华中科技大学 | A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm |
| CN106485247A (en) * | 2016-09-30 | 2017-03-08 | 广西师范大学 | Significance detection method based on neuron receptive field space structure |
| CN108520517A (en) * | 2018-04-10 | 2018-09-11 | 四川超影科技有限公司 | Method for detecting leakage based on machine vision |
| CN111179293A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic contour detection method based on color and gray level feature fusion |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246593A (en) * | 2008-03-27 | 2008-08-20 | 北京中星微电子有限公司 | Color image edge detection method and apparatus |
| CN101465001A (en) * | 2008-12-31 | 2009-06-24 | 昆山锐芯微电子有限公司 | Method for detecting image edge based on Bayer RGB |
| CN102163325A (en) * | 2011-04-06 | 2011-08-24 | 南京大学 | Edge feature detection method of multi-spectral image |
-
2011
- 2011-12-28 CN CN201110448119.1A patent/CN102567969B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246593A (en) * | 2008-03-27 | 2008-08-20 | 北京中星微电子有限公司 | Color image edge detection method and apparatus |
| CN101465001A (en) * | 2008-12-31 | 2009-06-24 | 昆山锐芯微电子有限公司 | Method for detecting image edge based on Bayer RGB |
| CN102163325A (en) * | 2011-04-06 | 2011-08-24 | 南京大学 | Edge feature detection method of multi-spectral image |
Non-Patent Citations (2)
| Title |
|---|
| 余见: "彩色图像边缘检测和分类", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 08, 31 August 2009 (2009-08-31) * |
| 郎方年 等: "四元数与彩色图像边缘检测", 《计算机科学》, vol. 34, no. 11, 31 December 2007 (2007-12-31) * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105139391A (en) * | 2015-08-17 | 2015-12-09 | 长安大学 | Edge detecting method for traffic image in fog-and-haze weather |
| CN105139391B (en) * | 2015-08-17 | 2018-01-30 | 长安大学 | A kind of haze weather traffic image edge detection method |
| CN105957067A (en) * | 2016-04-23 | 2016-09-21 | 北京工业大学 | Color image edge detection method based on color difference |
| CN105957067B (en) * | 2016-04-23 | 2018-10-19 | 北京工业大学 | A kind of color image edge detection method based on color difference |
| CN106228547A (en) * | 2016-07-15 | 2016-12-14 | 华中科技大学 | A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm |
| CN106228547B (en) * | 2016-07-15 | 2018-12-28 | 华中科技大学 | A kind of profile and border detection algorithm of view-based access control model color theory and homogeneity inhibition |
| CN106485247A (en) * | 2016-09-30 | 2017-03-08 | 广西师范大学 | Significance detection method based on neuron receptive field space structure |
| CN108520517A (en) * | 2018-04-10 | 2018-09-11 | 四川超影科技有限公司 | Method for detecting leakage based on machine vision |
| CN111179293A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic contour detection method based on color and gray level feature fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102567969B (en) | 2014-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102567969B (en) | Color image edge detection method | |
| CN104966085B (en) | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features | |
| CN103186904B (en) | Picture contour extraction method and device | |
| CN106650770A (en) | Mura defect detection method based on sample learning and human visual characteristics | |
| CN102750535B (en) | Method and system for automatically extracting image foreground | |
| CN105913093A (en) | Template matching method for character recognizing and processing | |
| CN103238335A (en) | Image processing device, image processing method, and program | |
| CN101246593B (en) | Color image edge detection method and apparatus | |
| CN101964293B (en) | Metallographical microstructural image processing method | |
| CN109035377B (en) | Image processing method for obtaining embossment sketch effect suitable for laser engraving | |
| CN107392880A (en) | A kind of imitative pattern painting automatic generation method | |
| CN106327537A (en) | Method and device for image pre-processing | |
| CN104217440B (en) | A kind of method extracting built-up areas from remote sensing images | |
| CN104504722A (en) | Method for correcting image colors through gray points | |
| CN106570839A (en) | Red Channel prior based underwater image sharpening method | |
| CN105913460A (en) | Skin color detection method and device | |
| Montseny et al. | A fuzzy approach to white blood cells segmentation in color bone marrow images | |
| CN103745231B (en) | Teleutospore image identification method for Tillctia Controversa Kahn (TCK) and allied variety TCT (Tilletia caries (DC.) Tul.) of TCK | |
| CN103559486B (en) | A kind of method of Video Detection floor positive and negative | |
| CN106780334A (en) | A kind of image classification method and system | |
| CN103714336A (en) | Biquaternion and echo state network-based fabric image edge detection method | |
| CN108389182A (en) | A kind of picture quality detection method and device based on deep neural network | |
| CN104463917B (en) | Based on the normalized image vision conspicuousness detection method of division | |
| CN109102482B (en) | Processing method for obtaining relief antique effect image suitable for laser engraving | |
| Kulshrestha et al. | Review of color blindness removal methods using image processing |
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 |