CN104599271B - CIE Lab color space based gray threshold segmentation method - Google Patents
CIE Lab color space based gray threshold segmentation method Download PDFInfo
- Publication number
- CN104599271B CN104599271B CN201510027820.4A CN201510027820A CN104599271B CN 104599271 B CN104599271 B CN 104599271B CN 201510027820 A CN201510027820 A CN 201510027820A CN 104599271 B CN104599271 B CN 104599271B
- Authority
- CN
- China
- Prior art keywords
- gray
- sigma
- cie lab
- threshold
- lab color
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Abstract
A CIE Lab color space based gray threshold segmentation method comprises the following steps of 1 transforming an image of a RGB color space into a CIE Lab color space, 2 conducting Gaussian histogram filtering on all gray channels of the CIE Lab color space, 3 adopting an Otsu threshold value method to calculate threshold values of the gray channels and adjusting the threshold values as local minimum, 4 calculating gray separation degrees of the gray channels of the CIE Lab color space, selecting the gray channel with highest gray separation degree and adopting the corresponding threshold value calculated in the step 3 to perform binaryzation division. By means of the CIE Lab color space based gray threshold segmentation method, the problem of division of a part of color images is well solved. In addition, the division operation can serve as pre-division operation conducted on complicated color images, and proposed separation degree index can also serve as a standard for image estimation.
Description
Technical field
The present invention relates to technical field of image processing, it is particularly a kind of based on color space to gray space change it is simple
Effective global threshold dividing method.
Background technology
Image segmentation is the classic problem in image procossing, and the matter of utmost importance of graphical analyses and pattern recognition.Image
Segmentation refers to and will make a distinction with the zones of different of special connotation in image, these regions be it is Uncrossed mutually, each
Region all meets the concordance of specific region.
Most research work all concentrates on the segmentation of gray level image at present, and the image partition method of main flow can be divided into
Three classes.One, based on the method for pixel, the also referred to as scheme based on threshold value, the classical scheme of comparison has peak valley method, minimum error
Method, maximum variance between clusters, maximum-entropy automatic threshold.Two, based on the method in region, have a region-growing method, regional split with
Merge and send out, watershed algorithm etc..Three, based on the method on border, there are various arithmetic operators Sobel gradients, Laplce's ladder
Degree operator etc..Four, based on the algorithm of particular theory, such as wavelet analysises, level cut, grab cut, ant group algorithm etc..By
It is limited to the identification ability of gray level in human eye, tens kinds of gray levels can only be recognized, but thousands of kinds of colors can be recognized.And
Coloured image compares gray level image there is provided more information, therefore also begins to be subject to more next based on the image segmentation of color space
More concerns, but relative to the research of gray level image segmentation, achievement is also far from enough.And in actual application, no matter
It is the image of camera or photographic head acquisition.Even medical science pseudo color image all not gray level images, so research cromogram
The segmentation of picture is with very real meaning.Although existing many ripe Segmentation of Color Image, such as Kmeans, watershed
Deng, but each own specific range of application, constantly open up the pursuit that new algorithm is researchers all the time.
Herein in this context, it is proposed that a kind of brand-new very simple color images scheme, it can be with
Split for common natural color image segmentation and medical science pseudo color image.
For many natural color images, especially plant and animal class image, as itself is in color or bright
There is good separation property relative to background on degree, now there is no need using complicated color images scheme.CIE Lab
L, a, b each passage can uniformly and independently reflect coloured image respectively in brightness, and, to green component, yellow is extremely for redness
Change on blue component, so as to and using grey relevant dynamic matrix complete color images provide may.For common ash
Degree medical image, is carried out Pseudocolor first, can project and strengthen the edge of tissue, and be converted to CIE again
The L of Lab color spaces, a, b passage, can compare original gray level image, preferably reflect soft tissue, and lesion tissue is relative
In the contrast of background, such that it is able to adopt basic threshold method to complete more accurate preextraction.
The content of the invention
It is an object of the present invention to propose a kind of gray level threshold segmentation method based on CIE Lab color spaces, the method
It is unsupervised, simple and effective color images scheme.It does not have the parameter of sensitivity, the iterative process for yet not having complexity, and
It is the brightness that make use of object and background, the segmentation of coloured image is converted into the segmentation of gray level image, so as to very by chrominance information
Good solves the problems, such as a part of color images.And the segmentation can be used as the pre-segmentation of complicated color image, institute
The separating degree index of proposition can also be used as picture appraisal standard.
The present invention provides a kind of gray level threshold segmentation method based on CIE Lab color spaces, comprises the steps:
Step 1:The image of RGB color space is converted into into CIE Lab color spaces;
Step 2:Gauss rectangular histogram filtering is carried out to each gray channel of CIE Lab color spaces;
Step 3:The threshold value of each gray channel is calculated using Otsu threshold methods, and threshold value is adjusted for local minimum;
Step 4:The gray scale separating degree of each gray channel of CIE Lab color spaces is calculated, gray scale separating degree is chosen most
Big gray channel, carries out binarization segmentation using the corresponding threshold value calculated in step 3.
From above-mentioned technical proposal as can be seen that the present invention has following technique effect:
1st, the gray level threshold segmentation method based on CIE Lab color spaces that the present invention is provided, it is proposed that a new figure
As evaluation criterion, i.e. gray level image separating degree index η, for the separable characteristic of the different gray level images of comparison.
2nd, the gray level threshold segmentation method based on CIE Lab color spaces that the present invention is provided, by the segmentation of coloured image
Problem, is converted to the segmentation problem of gray level image, reduces former problematic difficulty, in actual natural image segmentation and puppet
In the application of color medical image segmentation, good effect is achieved.
3rd, the gray level threshold segmentation method based on CIE Lab color spaces that the present invention is provided, for classical Otsu threshold values
Method can not obtain the shortcoming of effective threshold value when foreground and background variance is widely different, it is proposed that very simple and effective
Amendment scheme.
Description of the drawings
To further illustrate the technology contents of the present invention, below in conjunction with accompanying drawing and case study on implementation to the detailed description of the invention such as
Afterwards, wherein:
Fig. 1 be the present invention provide gray level threshold segmentation method based on CIE Lab color spaces the step of flow chart;
Fig. 2 is normalizing in separating degree in the gray level threshold segmentation scheme based on CIE Lab color spaces that the present invention is provided
Change the schematic diagram of area;
Fig. 3 is that the experiment sample RGB in case study on implementation of the present invention schemes and R, G, channel B figure;
Fig. 4 is that the experiment sample Lab in case study on implementation of the present invention schemes and L, a, b passage figure;
Fig. 5 is the segmentation result of part Experiment sample of the present invention, and segmentation result is shown by the way of superposition profile.
Fig. 6 is the RGB figures of cerebral tumor experiment sample of the present invention, corresponding Lab figures and corresponding gray-scale maps;
Fig. 7 is the corresponding L in CIE Lab color spaces of Fig. 6 midbrain tumors experiment samples, a, b gray channel figure;
Specific embodiment
Refer to shown in Fig. 1, the present invention provides a kind of gray level threshold segmentation method based on CIE Lab color spaces, bag
Include following steps:
Step 1:The image of RGB color space is converted into into CIE Lab color spaces, described RGB color space to CIE
Lab color space conversions, are changed to CIE XYZ colors space from rgb space using first, and reconvert is colored empty to CIE Lab
Between algorithm, specifically comprise the steps of;
Step 1a:Change in RGB to CIE XYZ colors space
Rgb color pattern is the color standard on the basis of physiology, based on the trichromatic industrial quarters of light.The three of light
Primary colors is red, green and blueness, and this is that have three kinds of cone cells most sensitive to red, green and blue light respectively due to the mankind.RGB color
Color pattern is by each to obtain to red (R), green (G), the change of blue (B) three Color Channels and their superpositions each other
The color of formula various kinds, this standard almost includes all colours that human eyesight can perceive, and is to use most wide at present
One of color system.
In the research of color-aware, CIE 1931XYZ color spaces (also referred to as 1931 color spaces of CIE) are wherein
One at first using mathematical way come the color space for defining, it by International Commission on Illumination (CIE) in 1931 found.Because
Human eye has the color sensor of the three types of response different wavelength range, and the complete drawing of all visible colors is three-dimensional
's.But the concept of color can be divided into two parts:Lightness and colourity.For example, white is bright color, and Lycoperdon polymorphum Vitt is considered as
It is less bright white.In other words, as white with the colourity of Lycoperdon polymorphum Vitt is, and lightness is different.
CIE XYZ color spaces design lightness or the measurement of brightness for causing Y parameter to be color, and the colourity of color is then led to
Cross derived parameter x and y to specify, they be the function of all three tristimulus value X, Y and Z in normalized three values two
It is individual.With reference to as follows for the 8-bit color image of electronic system, the conversion of RGB-XYZ in OpenCV:
Step 1b:CIE XYZ to CIE Lab color space conversions
Lab was set up on the basis of International Commission on Illumination (CIE) the color measurements international standard formulated in 1931
Come.1976, it is modified after be officially named CIE Lab.It is a kind of device-independent color system, is also a kind of base
In the color system of physiological feature.CIE Lab describe the visual response of people with method for digitizing, are devoted to perceiving uniformly
Property.L * component in Lab color spaces is used to represent the brightness of pixel that span to be [0,100], represents from black to pure white;
A represents the scope from redness to green, and span is [127, -128];B is represented from yellow to blueness
Scope, span is [127, -128].Can be accurate to do by changing the output levels of a and b components
Color balance, or using L * component adjusting luminance contrast.No direct conversion formula between RGB and LAB, which must use logical
Road XYZ color space is used as intermediate layer.With reference in OpenCV, for the 8-bit color image of electronic system, the conversion of XYZ-Lab is such as
Under:
X=X/0.950456
Z=Z/1.088754
A=500 (f (X)-f (Y))+128
B=200 (f (Y)-f (Z))+128
Wherein
The scope of resulting L, a, b is
0≤L≤100, -127≤a≤127, -127≤b≤127.
The scope of 8 can be converted it to by following formula
L=L × 255/128, a=a+128, b=b+128.
Why CIE Lab color spaces are adopted, be because that it compares RGB color space, with more preferable uniform illumination
Property.L * channel eliminates the impact of color, and a eliminates the impact of brightness with b passages, and tri- passages of R, G, B not only contain it is bright
The information of degree and color, even intercouples between three, not independent mutually.Refering to accompanying drawing 3 and 4, it can be seen that
In tri- passages of R, G, B, prospect pyramid and background sky are overlapped in gray space, and contrast is not high, it is difficult to adopt threshold
Value method makes a distinction.And in a passages, foreground and background then has good discrimination in gray space, accompanying drawing 5 is exactly to have selected
Prospect is opened with background separation by a passages using single threshold method, can be seen that segmentation result from the profile results being superimposed
Very accurately, i.e., foreground and background is very little in the overlap of the gray space of a passages.
Step 2:Gauss rectangular histogram filtering, described Gauss rectangular histogram are carried out to each passage of CIE Lab color spaces
Filtering, is, by Gaussian function, to carry out smothing filtering to grey level histogram, and specific algorithm is as follows:Hist (t) represents that gray level is
The probability density of t, gaussian filtering core variance of unit weight σ take 0.75;
After using the gaussian filtering, in rectangular histogram originally, many pseudo- peak values will be removed, follow-up so as to be conducive to
By the work that Otsu adjusting thresholds are local minimum.
Step 3:Described Otsu threshold methods, are a kind of methods for asking for global threshold.Threshold segmentation scheme is most ancient
And application most simply with universal method, its selection that it is critical only that threshold value is widely studied by Chinese scholars, non-
Often it is applied to the image that segmentation foreground and background has different grey-scale.Global thresholding can be divided into be examined based on histogram peak
The method of survey;Method based on Optimality Criteria such as maximum variance between clusters, are also Otsu methods, and it is by maximizing inter-class variance
To ask for threshold value;Maximum entropy method (MEM), it asks for threshold value by maximizing segmentation information entropy, and minimum cross entropy method is by minimizing segmentation
Intersection information entropy asks for threshold value, and minimum error rule asks for threshold value by minimizing Bayes risk;It is distributed based on spatial gradation
The method of information such as moment preserving method, asks for threshold value by causing segmentation figure picture and the moment preserving of source images;Based on domain transformation
The for example effective AVERAGE GRADIENT METHOD WITH of method;And threshold method of the method based on particular theory such as based on genetic algorithm etc..All of
In these methods, the maximum variance between clusters that Otsu is proposed, namely aforesaid Otsu methods show most stable, and without the need for ginseng
Number, for real world images maintain best uniformity and style characteristic, is adopted by business software GIMP and Academic Software Matlab
Receive as automatic threshold method, therefore be also adopted by the present invention.
Otsu threshold methods are, by grey level histogram, to calculate maximum variance between class, ask for threshold value t, and I is represented at gray level
Gray level image between [0,1 ..., L-1], ni represent the number of pixels that gray scale is n, and N represents the number of total pixel, so N
=Σ ni, the then Probability p of gray level niFor pi=ni/N.Assume that two classes are respectively C1And C2, split by threshold value t, C1Bag
Containing pixel of the gray level between [0 ..., t], C2Comprising pixel of the Pixel-level between [t+1 ..., L-1], p is made1(t)
And p2T () represents the cumulative probability of two classes, m1(t) and m2T () represents the gray average of two classes, σ2 1(t) and σ2 2T () represents two classes
Normalization variance, σ2 B(t) and σ2 wT () represents the inter-class variance and variance within clusters of whole image, σ2 T(t) representative image totality
Variance, then this tittle be calculated as follows:
Gray threshold t is obtained by formula (13);And described threshold value is adjusted so as to for local minimum,
It is that original rectangular histogram is divided into into two parts by t, corresponding peak value p can be obtained1And p2, then by calculating between peak value
Minimum valley t ' come the method that replaces original Otsu threshold values, the valley t ' for now obtaining better than original Otsu threshold values t because
Its cross point in two classes, uses it for Threshold segmentation, and two classes are in entropy of histogram by with minimum degree of overlapping.
Step 4:The gray scale separating degree of tri- gray channels of L, a, b of CIE Lab color spaces is calculated, is by self-defined
Separating degree criterion, it can characterize the degree of isolation characteristic of gray channel;Separating degree is bigger, and explanation prospect is divided with background
Property is stronger, less in the degree of overlapping of gray space, and separating degree is calculated as follows:Wherein mh、m1Highest and lowest gray value are represented,
s1、s2、s3The area Δ p of triangle is represented respectively1vp2、Δp2vv0With Δ p1vv0, details can refer to accompanying drawing 2, in calculating three
During angular area, gray scale is normalized to 0 to 1 scope;Point p1With p2Peak point is represented respectively, and v represents valley point, v0Generation
Subpoint of the table valley point on gray scale axle:
ηcd=(m2-m1)/(mh-ml) (14)
ηarea=S1/(S1+S2+S3) (16)
η=(ηov+ηcd)log2(1+ηarea) (17);
After calculating the separating degree of each gray channel, the image for selecting separating degree maximum, which has the contrast of maximum,
Therefore binarization segmentation can be carried out using Otsu threshold methods, obtains preferable effect.This invention is just for two class image segmentations
Problem, i.e. image only include two class target of foreground and background, during for comprising multiclass, can use the present invention's with iteration
Method, i.e., divide out by certain class successively;ηovIt is the ratio of inter-class variance and global variance, it reflects in certain degree
The separability of two classes in gray level image is gone out;Research shows that two classes are less in the overlap of gray space, then ηovIt is bigger, but instead
Often not so;ηcdWhat is reflected is the distance of the gray space of two classes, also reflects the separability of two classes to a certain extent,
But it and ηovEqually, be not separability sufficient condition, i.e., cannot be effectively reflected detached effectiveness, and ηareaThen
It is the sufficient condition of separability, i.e. ηareaBigger, separability is stronger, but ηareaRdativery sensitive, high ηareaIndex can not be made
For the essential condition of strong separability, therefore the characteristics of comprehensive three, it is proposed that η is used as final degree of isolation index, practice
Show, it is better than above-mentioned any single index therein, be very effective quantitative separating degree description.
Case study on implementation 1
The effect of gray level threshold segmentation scheme based on CIE Lab color spaces is provided for the checking present invention, carried out as
Lower experiment:Choose experiment sample as shown in Figure 5.Refering in RGB color image in accompanying drawing 3 and respective component, and accompanying drawing 4
Lab coloured images and respective component can be seen that:R originally, G, B component can not be well by foreground and background in ashes
Distinguish on degree figure, the effect of L * channel is similar to, and this explanation cannot effectively distinguish foreground and background by brightness.And a passages are then
Distinguish foreground and background in gray scale well, the respective gray scale separating degree of L, a, b passage is respectively 0.0989,
0.6871st, 0.4173, this can also reflect the effectiveness of separating degree, segmentation result as shown in Figure 5, the white contours of superposition
As segmentation result, accompanying drawing 5 also show the segmentation effect of some other images, and superposition white contours and red wheel is respectively adopted
Wide method is shown, it can be seen that this method is very effective.
Case study on implementation 2
This method can also be applied to medical image segmentation well, and accompanying drawing 6 shows pseudo-colourss RGB of a width cerebral tumor
Medical image, and corresponding Lab coloured images and gray level image, Fig. 7 show corresponding L, a, b gray level image.Contrast can
To find out, in a and b passages, the contrast of tumor and surrounding tissue is higher, and actually separating degree is also that b passages are maximum, so as to can
To compare gray-scale maps, more preferable threshold application method carries out pre-segmentation, more accurately to extract pathological tissues.
Claims (6)
1. a kind of gray level threshold segmentation method based on CIE Lab color spaces, comprises the steps:
Step 1:The image of RGB color space is converted into into CIE Lab color spaces;
Step 2:Gauss rectangular histogram filtering is carried out to each gray channel of CIE Lab color spaces;
Step 3:The threshold value of each gray channel is calculated using Otsu threshold methods, and threshold value is adjusted for local minimum;
Step 4:The gray scale separating degree of each gray channel of CIE Lab color spaces is calculated, gray scale separating degree maximum is chosen
Gray channel, carries out binarization segmentation using the corresponding threshold value calculated in step 3;
Wherein described gray scale separating degree, is self-defining separating degree criterion, and the degree of isolation that it can characterize gray channel is special
Property;Separating degree is bigger, illustrates that prospect is stronger with the separability of background, the calculating of separating degree less in the degree of overlapping of gray space
It is as follows:Wherein mh, m1Represent highest and lowest gray value, m1、m2Represent the average gray value of two classes, σ2 BT () represents side between class
Difference, σ2 TRepresentative image variance, s1、s2、s3The area Δ p of triangle is represented respectively1vp2、Δp2vv0With Δ p1vv0, calculating three
During angular area, gray scale is normalized to 0 to 1 scope;Point p1With p2Peak point is represented respectively, and v represents valley point, v0Generation
Subpoint of the table valley point on gray scale axle:
ηcd=(m2-m1)/(mh-ml) (14)
ηarea=S1/(S1+S2+S3) (16)
η=(ηov+ηcd)log2(1+ηarea) (17)。
2. the gray level threshold segmentation method based on CIE Lab color spaces as claimed in claim 1, wherein described RGB is color
The colour space is changed to CIE Lab spaces, is changed to CIE XYZ space from rgb space using first, and reconvert is empty to CIE Lab
Between algorithm.
3. the gray level threshold segmentation method based on CIE Lab color spaces as claimed in claim 1, wherein described Gauss is straight
Side's figure filtering, is, by Gaussian function, to carry out smothing filtering to grey level histogram, and specific algorithm is as follows:Hist (t) represents gray scale
Probability density of the level for t, gaussian filtering kernel function KσVariance of unit weight σ takes 0.75;
4. the gray level threshold segmentation method based on CIE Lab color spaces as claimed in claim 1, wherein described Otsu thresholds
Value method, is, by grey level histogram, to calculate maximum variance between class, asks for threshold value t, and I represents gray level in [0,1 ..., L-
1] gray level image between, ni represent the number of pixels that gray scale is n, and N represents the number of total pixel, so N=∑ ni, then gray scale
The Probability p of level niFor pi=ni/N;Assume that two classes are respectively C1And C2, split by threshold value t, C1It is in comprising gray level
Pixel between [0 ..., t], C2Comprising pixel of the Pixel-level between [t+1 ..., L-1], p is made1(t) and p2T () represents two
The cumulative probability of class, m1(t) and m2T () represents the gray average of two classes, σ2 1(t) and σ2 2T () represents the normalization variance of two classes,
σ2 B(t) and σ2 wT () represents the inter-class variance and variance within clusters of whole image, σ2 TRepresentative image variance, then the calculating of this tittle is such as
Under:
Gray threshold t is obtained by formula (13).
5. the gray level threshold segmentation method based on CIE Lab color spaces as claimed in claim 1, wherein described adjustment threshold
Value, is to obtain initial threshold t by 0tsu threshold methods are applied to rectangular histogram, and original rectangular histogram is divided into two parts by t, can be with
Obtain corresponding peak value p1And p2, then by calculating the minimum valley t ' between peak value replacing the side of original Otsu threshold values
Method, the valley t ' for now obtaining is better than original Otsu threshold values t.
6. the gray level threshold segmentation method based on CIE Lab color spaces as claimed in claim 1, wherein described binaryzation
Segmentation, refers to gray scale separating degree maximum gray channel, according to corresponding threshold value, gray value is classified as one more than threshold value
Class, less than threshold value be classified as it is another kind of.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510027820.4A CN104599271B (en) | 2015-01-20 | 2015-01-20 | CIE Lab color space based gray threshold segmentation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510027820.4A CN104599271B (en) | 2015-01-20 | 2015-01-20 | CIE Lab color space based gray threshold segmentation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104599271A CN104599271A (en) | 2015-05-06 |
CN104599271B true CN104599271B (en) | 2017-04-12 |
Family
ID=53125021
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510027820.4A Active CN104599271B (en) | 2015-01-20 | 2015-01-20 | CIE Lab color space based gray threshold segmentation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104599271B (en) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105678310B (en) * | 2016-02-03 | 2019-08-06 | 北京京东方多媒体科技有限公司 | Thermal-induced imagery contour extraction method and device |
CN107204001B (en) * | 2016-03-16 | 2020-03-31 | 飞依诺科技(苏州)有限公司 | Automatic midium segmentation method in carotid artery ultrasonic image |
CN106023143B (en) * | 2016-05-06 | 2018-11-13 | 中铁大桥科学研究院有限公司 | Wave height monitoring and pre-alarming method at a kind of sea-spanning bridge con struction |
CN106023215A (en) * | 2016-05-24 | 2016-10-12 | 北京农业智能装备技术研究中心 | Method and system for distinguishing field crops from background |
CN107438178B (en) * | 2016-05-25 | 2019-02-22 | 掌赢信息科技(上海)有限公司 | A kind of image color antidote and electronic equipment |
CN106374394A (en) * | 2016-09-28 | 2017-02-01 | 刘子轩 | Pipeline robot based on image recognition technology and control method |
CN106546756B (en) * | 2016-10-20 | 2018-10-26 | 浙江孕橙医疗科技有限公司 | A kind of ovulation test paper detection method and system |
CN107274426B (en) * | 2017-06-22 | 2023-08-18 | 中国农业机械化科学研究院 | On-line detection method and system for field straw coverage rate |
CN108921868B (en) * | 2018-07-02 | 2021-08-24 | 中央民族大学 | Improved Otsu threshold segmentation method |
CN109214996B (en) * | 2018-08-29 | 2021-11-12 | 深圳市元征科技股份有限公司 | Image processing method and device |
CN110706183B (en) * | 2019-10-11 | 2023-05-19 | 成都极米科技股份有限公司 | Image definition determining method and device, projector device and storage medium |
WO2021155438A1 (en) * | 2020-02-06 | 2021-08-12 | Canva Pty Ltd | Systems and methods for automatically adjusting design element attributes |
CN115439509B (en) * | 2022-11-07 | 2023-02-03 | 成都泰盟软件有限公司 | Multi-target tracking method and device, computer equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101964108A (en) * | 2010-09-10 | 2011-02-02 | 中国农业大学 | Real-time on-line system-based field leaf image edge extraction method and system |
CN102436644A (en) * | 2011-11-02 | 2012-05-02 | 南京物联网研究院发展有限公司 | Unstructured road detection method based on adaptive edge registration |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7248736B2 (en) * | 2004-04-19 | 2007-07-24 | The Trustees Of Columbia University In The City Of New York | Enhancing images superimposed on uneven or partially obscured background |
US20120277571A1 (en) * | 2011-04-26 | 2012-11-01 | Korea Basic Science Institute | Method For Measuring Trabecular Bone Parameters From MRI Images |
-
2015
- 2015-01-20 CN CN201510027820.4A patent/CN104599271B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101964108A (en) * | 2010-09-10 | 2011-02-02 | 中国农业大学 | Real-time on-line system-based field leaf image edge extraction method and system |
CN102436644A (en) * | 2011-11-02 | 2012-05-02 | 南京物联网研究院发展有限公司 | Unstructured road detection method based on adaptive edge registration |
Non-Patent Citations (2)
Title |
---|
基于图像不确定性信息的阈值分割方法研究;雷博;《中国博士学位论文全文数据库.信息科技辑》;20140415(第4期);摘要第2段、第3页第3段 * |
基于彩色与深度图像的唇读技术研究;孙威;《中国优秀硕士学位论文全文数据库.信息科技辑》;20140115(第1期);第33页第3段、34页第3段、35页第4段、36页第3段 * |
Also Published As
Publication number | Publication date |
---|---|
CN104599271A (en) | 2015-05-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104599271B (en) | CIE Lab color space based gray threshold segmentation method | |
CN103247059B (en) | A kind of remote sensing images region of interest detection method based on integer wavelet and visual signature | |
EP3300024B1 (en) | Color identification system, color identification method, and display device | |
CN106388781A (en) | Method for detecting skin colors and pigmentation situation of skin | |
Krawczyk et al. | Lightness perception in tone reproduction for high dynamic range images | |
CN103914699A (en) | Automatic lip gloss image enhancement method based on color space | |
Rosenthal et al. | Color statistics of objects, and color tuning of object cortex in macaque monkey | |
CN105893925A (en) | Human hand detection method based on complexion and device | |
CN104598908A (en) | Method for recognizing diseases of crop leaves | |
Berezhnoy et al. | Computer analysis of van Gogh’s complementary colours | |
Casanova et al. | Texture analysis using fractal descriptors estimated by the mutual interference of color channels | |
Cirillo et al. | Tensor decomposition for colour image segmentation of burn wounds | |
Zakir et al. | Road sign segmentation based on colour spaces: A Comparative Study | |
US8249342B1 (en) | Color analytics for a digital image | |
Wicaksono et al. | Color and texture feature extraction using gabor filter-local binary patterns for image segmentation with fuzzy C-means | |
Hussain et al. | Color constancy algorithm for mixed-illuminant scene images | |
CN113129390B (en) | Color blindness image re-coloring method and system based on joint significance | |
CN111160194B (en) | Static gesture image recognition method based on multi-feature fusion | |
CN108364011A (en) | PolSAR image multi-stage characteristics extract and unsupervised segmentation method | |
Chen et al. | Automated bridge coating defect recognition using adaptive ellipse approach | |
Liu et al. | Color constancy by combining low-mid-high level image cues | |
DE19838806A1 (en) | Method and device for detecting object colors | |
CN106204476A (en) | A kind of low-luminance color image enchancing method | |
Choukikar et al. | Segmenting the Optic Disc in retinal images using bi-histogram equalization and thresholding the connected regions | |
CN110097070B (en) | Chinese painting characteristic color set acquisition method based on human visual perception |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |