CN103955952A - Extraction and description method for garment image color features - Google Patents
Extraction and description method for garment image color features Download PDFInfo
- Publication number
- CN103955952A CN103955952A CN201410184295.2A CN201410184295A CN103955952A CN 103955952 A CN103955952 A CN 103955952A CN 201410184295 A CN201410184295 A CN 201410184295A CN 103955952 A CN103955952 A CN 103955952A
- Authority
- CN
- China
- Prior art keywords
- image
- color
- clothing
- main color
- main
- 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
Abstract
The invention discloses an extraction and description method for garment image color features. The method comprises the steps of firstly, conducting foreground image extraction and quantification on each image in a garment image training set and obtaining main colors; then dividing each foreground image into a plurality of local parts to obtain the main color of each local part, and conducting illumination normalization processing; finally clustering the main colors of all the local parts in the garment image training set to obtain K clusters, namely a code book. By the adoption of the same method, the number of the local parts in each cluster is counted, and codes of the garment images are formed. According to the extraction and description method for the garment image color features, the color (vision) features of interesting garment regions in the garment images can be effectively extracted and described, and the garment image colors are precisely and effectively described, so that feature description complexity is reduced, and precision and efficiency of garment image retrieval are improved.
Description
Technical field
The invention belongs to digital image processing techniques field, more specifically say, relate to a kind of extraction and describing method of image of clothing color characteristic.
Background technology
Along with computer process ability strengthens day by day, the improving constantly of the extensively universal and network bandwidth of technique of internet, a large amount of image of clothing information constantly produce.How from the image of clothing data of these magnanimity, searching for the interested image of clothing of people and effectively utilize these images, becoming problem in the urgent need to address, content-based image of clothing retrieval technique becomes one of major technique addressing this problem.
Content-based image of clothing retrieval technique, by extraction and description to the feature of image of clothing (color characteristic, textural characteristics, shape facility, spatial relationship feature etc.), reaches the object of identical or similar image of clothing retrieval.
Color characteristic, as a kind of global characteristics, has been described the surface nature of image or the corresponding object of image-region.General color characteristic is all the feature based on pixel, and now all pixels that belong to image or image-region have contribution separately.Because color is insensitive for the variations such as direction, size and visual angle of image or image-region, so color characteristic is widely used as the very strong visual signature of a kind of robustness.
The method of extracting color characteristic has a lot, wherein color histogram is the method for the most frequently used expression color characteristic, its advantage is the impact that not changed by image transfer and translation, the impact that further not changed by graphical rule by normalization, its shortcoming is not give expression to the information that color space distributes, and both cannot in Description Image, obtain a certain concrete object or object.Be directed to image of clothing color feature, traditional color feature method cannot be made the description becoming more meticulous to our interested clothes region, can not meet the requirement of high-accuracy for large-scale data retrieval.
So, for extraction and the describing method of image of clothing color characteristic, solve following problem: 1) get the interested clothes area image of people in image of clothing; 2) propose a kind of the color of interested clothes area image to be carried out to effective describing method; 3) color characteristic has low-dimensional, improves the recall precision of large-scale data.
1, prior art one related to the present invention
The Chinese invention patent application that on May 26th, 2010 announces, publication No. is 101714257A " method of the main color characteristic extraction of image and structural description " discloses the method for the main color characteristic extraction of a kind of image and structural description, comprises the following steps:
(a), input pending image;
(b), select color space;
(c), in color space, color of image is quantized;
(d) number of pixels that, calculates normalized color histogram and obtain every kind of color accounts for whole image pixel number object number percent;
(e), according to number percent, obtain the main color of image;
(f), calculate main color variance and spatial coherence;
(g), the main color characteristic of image is encoded and obtained the description about the main color characteristic of image.Shortcoming:
This invention is by extracting and structural description accounting for the main color of image certain percentage.But in fact this invention has following three shortcomings: be 1) statistics to image overall color, color extraction do not carried out in target area; 2) need to calculate the variance of every kind of main color, calculated amount lowers efficiency greatly; 3) adopt structurized description, for large-scale image, be difficult to carry out index foundation and similarity matching, cause recall precision low, accuracy rate can not guarantee.
2, prior art two related to the present invention
The Chinese invention patent " a kind of image search method based on multilayer figure color characteristic " that on November 28th, 2012 announces, publication No. is 101989302B discloses a kind of image search method based on multilayer figure color characteristic, comprising:
1), gridding is carried out in the color image space of query image, add up pixel number in each grid, choose the grid with local maximum;
2), in K means clustering algorithm by application a kind of new distance optimized algorithm and ENNS algorithm, generate fast each color cluster and reasonable statistical distribution center thereof;
3), on the other hand, query image is carried out to space partition, calculate the color average of its Gauss's weighting;
4), by the comparison at image subblock color average and the reasonable statistical distribution of color cluster center, extraction K layer bitmap feature;
5), the similarity measurement of Color cluster reasonable statistical distribution center and bitmap, carry out the coupling retrieval of characteristics of image.
Shortcoming:
This invention, by a kind of color characteristic based on multilayer bitmap, realizes image retrieval.Although but in fact this invention obtains comparatively comprehensively extracting and describing to color of image, still has significant limitation in costume retrieval.This invention, using color characteristic as the global characteristics of describing entire image, is not carried out effective background elimination, fails to ignore the impact that background produces color of image feature, does not obtain the description of the object that in image, people are concerned about most; Have, the feature obtaining by this kind of method is described complicated, the similarity matching between effect characteristics and image retrieval accuracy rate again.
Summary of the invention
The object of the invention is to overcome the deficiencies in the prior art, a kind of extraction and describing method of image of clothing color characteristic are provided, accurately and efficiently image of clothing color is being described simultaneously, the complexity of describing to reduce feature, improves accuracy and the efficiency of image of clothing retrieval.
For realizing above object, the extraction of image of clothing color characteristic of the present invention and describing method, is characterized in that, comprises the following steps:
(1), input pending image of clothing training set;
(2), from image of clothing training set, choose successively image of clothing and carry out following processing:
2.1), extract the foreground image of image of clothing;
2.2), the color of foreground image is carried out to quantification treatment according to color space;
2.3), obtain the main color of foreground image:
A), calculate the normalized color histogram of foreground image, obtain every kind of color and account for whole image pixel number object number percent;
B), according to number percent, obtain a plurality of main color of foreground image: set a percentage threshold T, when certain color accounts for whole image pixel number object number percent and is greater than percentage threshold T, think that this color is the main color of foreground image;
2.4), foreground image be divided into several localized masses and calculate the main color under each localized mass: the color to localized mass averages, each main color of the mean value obtaining and foreground image compares, when difference is all greater than setting value, think that this localized mass does not have corresponding main color, is designated as 0 value; When if difference is less than setting value, the main color using the main color of difference minimum as this localized mass;
2.5), the main color of each localized mass is carried out unitary of illumination processing;
(3), the main color of all localized masses of all image of clothing in image of clothing training set is carried out to cluster, obtain K cluster, i.e. code book;
(4), according to code book, image of clothing is encoded:
The image of clothing of needs coding is processed according to the method for step (2), obtain the main color of each localized mass of image of clothing, then the code book obtaining according to step (3) carries out merge sort, by the main color of each localized mass foundation, by its merger, it is a certain cluster, finally count the quantity of the localized mass of each cluster, and form the coding of this image of clothing.
The object of the present invention is achieved like this:
The extraction of image of clothing color characteristic of the present invention and describing method, first by each image to image of clothing training set, carry out foreground image extraction, quantize and obtain main color, obtain a plurality of main colors, then foreground image is divided into several localized masses, obtain the main color of each localized mass, and carry out unitary of illumination processing, and finally the main color of all localized masses of all image of clothing in image of clothing training set is carried out to cluster, obtaining K cluster is code book.Like this when image of clothing is described, the main color of each localized mass that adopting uses the same method extracts image of clothing, then according to code book by each localized mass according to main color, by its merger, it is a certain cluster, finally count the quantity of the localized mass of each cluster, and form the coding of this image of clothing.
The extraction of image of clothing color characteristic of the present invention and describing method are owing to can color (vision) feature in interested clothes region in image of clothing effectively being extracted and be described, accurately and efficiently image of clothing color is described simultaneously, the complexity of describing to reduce feature, improves accuracy and the efficiency of image of clothing retrieval.Therefore when image of clothing is retrieved, can judge by the similarity between the color characteristic of comparison two width image of clothing the similarity of two width image of clothing, thereby can arrive identical or similar image by fast search.Method of the present invention not only processing procedure is simple and convenient, and stable and reliable working performance, raising image of clothing retrieval rate, enhancing are had to important effect to the analysis of image of clothing and understandability, can be general for fields such as image of clothing retrieval, machine vision and interactive videos.
Accompanying drawing explanation
Fig. 1 is the extraction of image of clothing color characteristic of the present invention and the process flow diagram of describing method;
Fig. 2 is the image of clothing that need to carry out color characteristic extraction;
Fig. 3 is the foreground image of image of clothing shown in Fig. 2 after extracting;
Fig. 4 is the result figure that image of clothing shown in Fig. 2 is retrieved in 6627 image of clothing test sets.
Embodiment
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described, so that those skilled in the art understands the present invention better.Requiring particular attention is that, in the following description, when perhaps the detailed description of known function and design can desalinate main contents of the present invention, these are described in here and will be left in the basket.
In the present embodiment, as shown in Figure 1, extraction and the describing method of image of clothing color characteristic of the present invention comprise the steps:
1, input pending image of clothing training set
In specific implementation process, in image of clothing training set, the color space of image of clothing can be rgb space, YCbCr space, HSV space, HMMD space or monochromatic space.In the present embodiment, select rgb space, in image of clothing training set, the quantity of image of clothing is 6627.
2, in the present embodiment, the foreground image of extraction image of clothing specifically comprises the following steps:
(a), the foreground area of image of clothing is set;
(b), adopting Grabcut method to carry out image to image of clothing cuts apart, obtains interested image of clothing, i.e. foreground image.
Fig. 2 is the image of clothing that need to carry out color characteristic extraction, and the resolution of image of clothing is 412 * 391, and picture format is JPEG, and the foreground image after extraction as shown in Figure 3.
3, the color of foreground image is carried out to quantification treatment according to color space, specifically comprises the following steps:
(a), color space is divided into several little color spaces, each little color space claims one dimension;
(b), the color of each pixel of this image of clothing is quantized according to described dimension.
In the present embodiment, adopt rgb space, each pixel color of its image of clothing is quantized respectively according to RGB.In the present embodiment, each color component of pixel is 0~255, is quantified as 0~16 to reduce the treatment capacity of data, is quantified as 16.
4, in the present embodiment, calculate the normalized color histogram of foreground image, obtain every kind of color and account for whole image pixel number object number percent, be specially:
Pi=Mi/M;
Wherein, Mi represents the pixel quantity of i kind color in image of clothing, and M represents the pixel quantity of whole image.
According to number percent Pi, obtain a plurality of main color of foreground image, in the present embodiment, percentage threshold T is 0.8%, when certain color i accounts for whole image pixel number object number percent and is greater than percentage threshold 0.8%, thinks that this color is the main color of foreground image.
In the present embodiment, obtain 15 main colors, corresponding RGB component is respectively:
(11,9,10),(12,10,10),(12,10,11),(12,11,11),(12,12,12),(13,10,11),(13,11,11),(13,11,12),(13,12,12),(14,11,12),(14,12,12),(14,12,13),(14,13,13),(14,14,14),(15,15,15)。
5, foreground image be divided into several localized masses and calculate the main color under each localized mass: the color to localized mass averages, each main color of the mean value obtaining and foreground image compares, when difference is all greater than setting value, think that this localized mass does not have corresponding main color, is designated as 0 value; When if difference is less than setting value, the main color using the main color of difference minimum as this localized mass.
In this enforcement, foreground image is divided into 14 * 14 localized masses, travel through each localized mass, the dominant color set that obtains each localized mass the best shows, above-mentioned 15 kinds of main colors is a kind of, if localized mass does not have corresponding main color, its main color is set to (0,0,0), finally obtaining 14 * 14 is the main color of 196 localized masses.
6, the main color of each localized mass is carried out unitary of illumination processing, is specially:
According to following formula, carry out unitary of illumination processing:
(x,y,z)=(X/(X+Y+Z),Y/(X+Y+Z),Z/(X+Y+Z))
Wherein, X, Y and Z represent the unitary of illumination main color RGB component value of localized mass before, and x, y and z represent the main color RGB component value after unitary of illumination.
7, the main color of all localized masses of all image of clothing in image of clothing training set is carried out to cluster, obtain K cluster, i.e. code book.
In the present embodiment, adopt K-means cluster, obtain 65 clusters.
8, according to code book, image of clothing is encoded:
The image of clothing of needs coding is processed according to the method for step (2), obtain the main color of each localized mass of image of clothing, then the code book obtaining according to step (3) carries out merge sort, by the main color of each localized mass foundation, by its merger, it is a certain cluster, finally count the quantity of the localized mass of each cluster, and form the coding of this image of clothing, obtain the K dimension coding of image of clothing color characteristic.
In the middle of practical application, image of clothing color characteristic is exactly main color merger and the classification to the localized mass of image of clothing according to code book, obtains K dimension coding:
F={h
1,h
2,h
3,…,h
i,……,h
K};
Wherein, K represents cluster centre number, h
ithe number that represents the image of clothing localized mass of main color and i cluster centre.
Finally, according to cluster centre, the main color of the localized mass of the foreground image shown in Fig. 3 is carried out to merger and classification, produce the description about image of clothing color characteristic, the 65 dimension color characteristics that obtain the image of clothing shown in Fig. 2 encode, specific as follows:
F=(87,0,0,4,0,0,0,0,24,0,0,0,0,0,0,0,0,0,0,0,3,0,0,0,0,0,0,0,0,0,0,7,0,0,0,0,0,61,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,10,0,0,0,0,0)
Finally using the image of clothing shown in Fig. 2 as target image, using 6627 images in image of clothing training set as test set, that utilizes image of clothing is above-mentionedly coded in 6627 image of clothing and retrieves, and its result as shown in Figure 4.
In the present embodiment, use KL apart from the similarity of weighing two image of clothing color characteristics.
Although above the illustrative embodiment of the present invention is described; so that those skilled in the art understand the present invention; but should be clear; the invention is not restricted to the scope of embodiment; to those skilled in the art; as long as various variations appended claim limit and definite the spirit and scope of the present invention in, these variations are apparent, all utilize innovation and creation that the present invention conceives all at the row of protection.
Claims (3)
1. the extraction of image of clothing color characteristic and a describing method, is characterized in that, comprises the following steps:
(1), input pending image of clothing training set;
(2), from image of clothing training set, choose successively image of clothing and carry out following processing:
2.1), extract the foreground image of image of clothing;
2.2), the color of foreground image is carried out to quantification treatment according to color space;
2.3), obtain the main color of foreground image:
A), calculate the normalized color histogram of foreground image, obtain every kind of color and account for whole image pixel number object number percent;
B), according to number percent, obtain a plurality of main color of foreground image: set a percentage threshold T, when certain color accounts for whole image pixel number object number percent, be greater than percentage threshold T, think that this color is the main color of foreground image;
2.4), foreground image be divided into several localized masses and calculate the main color under each localized mass: the color to localized mass averages, each main color of the mean value obtaining and foreground image compares, when difference is all greater than setting value, think that this localized mass does not have corresponding main color, is designated as 0 value; When if difference is less than setting value, the main color using the main color of difference minimum as this localized mass;
2.5), the main color of each localized mass is carried out unitary of illumination processing;
(3), the main color of all localized masses of all image of clothing in image of clothing training set is carried out to cluster, obtain K cluster;
(4), according to code book, image of clothing is encoded:
The image of clothing of needs coding is processed according to the method for step (2), obtain the main color of each localized mass of image of clothing, then the code book obtaining according to step (3) carries out merge sort, by the main color of individual localized mass foundation, by its merger, it is a certain cluster, finally count the quantity of the localized mass of each cluster, and form the coding of this image of clothing.
2. the extraction of color characteristic according to claim 1 and describing method, is characterized in that, described color space can be rgb space.
3. the extraction of color characteristic according to claim 1 and describing method, is characterized in that, described percentage threshold T is 0.8%.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410184295.2A CN103955952A (en) | 2014-05-04 | 2014-05-04 | Extraction and description method for garment image color features |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410184295.2A CN103955952A (en) | 2014-05-04 | 2014-05-04 | Extraction and description method for garment image color features |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103955952A true CN103955952A (en) | 2014-07-30 |
Family
ID=51333221
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410184295.2A Pending CN103955952A (en) | 2014-05-04 | 2014-05-04 | Extraction and description method for garment image color features |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103955952A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268544A (en) * | 2014-10-14 | 2015-01-07 | 陶晨 | Evaluation system for clothes visual effect |
CN104484450A (en) * | 2014-12-25 | 2015-04-01 | 广东欧珀移动通信有限公司 | Clothing matching recommendation method and clothing matching recommendation device based on pictures |
CN104504734A (en) * | 2014-09-16 | 2015-04-08 | 浙江工业大学 | Image color transferring method based on semantics |
CN106227827A (en) * | 2016-07-25 | 2016-12-14 | 华南师范大学 | Image of clothing foreground color feature extracting method and costume retrieval method and system |
CN107153670A (en) * | 2017-01-23 | 2017-09-12 | 合肥麟图信息科技有限公司 | The video retrieval method and system merged based on multiple image |
WO2020151529A1 (en) * | 2019-01-23 | 2020-07-30 | 阿里巴巴集团控股有限公司 | Clothing image classification and image classification methods and apparatuses, and device |
CN111488885A (en) * | 2020-06-28 | 2020-08-04 | 成都四方伟业软件股份有限公司 | Intelligent extraction method and device for theme color system of picture |
CN111612011A (en) * | 2020-05-21 | 2020-09-01 | 杭州智珺智能科技有限公司 | Clothing color extraction method based on human body semantic segmentation |
CN111652859A (en) * | 2020-05-25 | 2020-09-11 | 杭州电子科技大学 | K-means-based cloth pattern intelligent color separation method |
CN112489142A (en) * | 2020-11-30 | 2021-03-12 | 济南博观智能科技有限公司 | Color identification method, device, equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080089615A1 (en) * | 2006-10-13 | 2008-04-17 | Canon Kabushiki Kaisha | Image searching apparatus, control method thereof, program, and storage medium |
CN101231757A (en) * | 2007-12-07 | 2008-07-30 | 北京搜狗科技发展有限公司 | Apparatus and method for analyzing picture dominant hue as well as application in picture searching thereof |
CN101714257A (en) * | 2009-12-23 | 2010-05-26 | 公安部第三研究所 | Method for main color feature extraction and structuring description of images |
-
2014
- 2014-05-04 CN CN201410184295.2A patent/CN103955952A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080089615A1 (en) * | 2006-10-13 | 2008-04-17 | Canon Kabushiki Kaisha | Image searching apparatus, control method thereof, program, and storage medium |
CN101231757A (en) * | 2007-12-07 | 2008-07-30 | 北京搜狗科技发展有限公司 | Apparatus and method for analyzing picture dominant hue as well as application in picture searching thereof |
CN101714257A (en) * | 2009-12-23 | 2010-05-26 | 公安部第三研究所 | Method for main color feature extraction and structuring description of images |
Non-Patent Citations (1)
Title |
---|
XIANWANG WANG AND TONG ZHANG: ""Clothes Search in Consumer Photos via Color Matching and Attribute Learning"", 《PROCEEDINGS OF THE 19TH ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA.ACM》 * |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104504734A (en) * | 2014-09-16 | 2015-04-08 | 浙江工业大学 | Image color transferring method based on semantics |
CN104504734B (en) * | 2014-09-16 | 2017-09-26 | 浙江工业大学 | A kind of color of image transmission method based on semanteme |
CN104268544A (en) * | 2014-10-14 | 2015-01-07 | 陶晨 | Evaluation system for clothes visual effect |
CN104268544B (en) * | 2014-10-14 | 2018-05-08 | 浙江出入境检验检疫局检验检疫技术中心 | A kind of clothes visual effect evaluation system |
CN104484450A (en) * | 2014-12-25 | 2015-04-01 | 广东欧珀移动通信有限公司 | Clothing matching recommendation method and clothing matching recommendation device based on pictures |
CN104484450B (en) * | 2014-12-25 | 2018-05-04 | 广东欧珀移动通信有限公司 | Clothing matching based on image recommends method and clothing matching recommendation apparatus |
CN106227827B (en) * | 2016-07-25 | 2019-04-12 | 华南师范大学 | Image of clothing foreground color feature extracting method and costume retrieval method and system |
CN106227827A (en) * | 2016-07-25 | 2016-12-14 | 华南师范大学 | Image of clothing foreground color feature extracting method and costume retrieval method and system |
CN107153670A (en) * | 2017-01-23 | 2017-09-12 | 合肥麟图信息科技有限公司 | The video retrieval method and system merged based on multiple image |
WO2020151529A1 (en) * | 2019-01-23 | 2020-07-30 | 阿里巴巴集团控股有限公司 | Clothing image classification and image classification methods and apparatuses, and device |
CN111476253A (en) * | 2019-01-23 | 2020-07-31 | 阿里巴巴集团控股有限公司 | Clothing image classification method, clothing image classification device, clothing image classification method, clothing image classification device and clothing image classification equipment |
CN111476253B (en) * | 2019-01-23 | 2024-04-02 | 阿里巴巴集团控股有限公司 | Clothing image classification method, device and equipment and image classification method and device |
CN111612011A (en) * | 2020-05-21 | 2020-09-01 | 杭州智珺智能科技有限公司 | Clothing color extraction method based on human body semantic segmentation |
CN111652859A (en) * | 2020-05-25 | 2020-09-11 | 杭州电子科技大学 | K-means-based cloth pattern intelligent color separation method |
CN111652859B (en) * | 2020-05-25 | 2023-11-07 | 杭州电子科技大学 | K-means-based intelligent color separation method for cloth patterns |
CN111488885A (en) * | 2020-06-28 | 2020-08-04 | 成都四方伟业软件股份有限公司 | Intelligent extraction method and device for theme color system of picture |
CN111488885B (en) * | 2020-06-28 | 2020-09-25 | 成都四方伟业软件股份有限公司 | Intelligent extraction method and device for theme color system of picture |
CN112489142A (en) * | 2020-11-30 | 2021-03-12 | 济南博观智能科技有限公司 | Color identification method, device, equipment and storage medium |
CN112489142B (en) * | 2020-11-30 | 2024-04-09 | 济南博观智能科技有限公司 | Color recognition method, device, equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103955952A (en) | Extraction and description method for garment image color features | |
CN102819582B (en) | Quick searching method for mass images | |
CN104599275B (en) | The RGB-D scene understanding methods of imparametrization based on probability graph model | |
DeGol et al. | Geometry-informed material recognition | |
Fan et al. | Unsupervised feature learning for land-use scene recognition | |
CN104112018B (en) | A kind of large-scale image search method | |
CN107122712B (en) | Palm print image identification method based on CNN and bidirectional VLAD | |
CN106126585B (en) | The unmanned plane image search method combined based on quality grading with perceived hash characteristics | |
CN101714257A (en) | Method for main color feature extraction and structuring description of images | |
CN104778701A (en) | Local image describing method based on RGB-D sensor | |
CN104361096B (en) | The image search method of a kind of feature based rich region set | |
CN108629783A (en) | Image partition method, system and medium based on the search of characteristics of image density peaks | |
Kavitha et al. | An efficient content based image retrieval using color and texture of image sub blocks | |
CN106203448B (en) | A kind of scene classification method based on Nonlinear Scale Space Theory | |
CN103942778A (en) | Fast video key frame extraction method of principal component characteristic curve analysis | |
CN104376312A (en) | Face recognition method based on word bag compressed sensing feature extraction | |
Lu et al. | Spectral segmentation via midlevel cues integrating geodesic and intensity | |
CN107358244B (en) | A kind of quick local invariant feature extracts and description method | |
CN106250828B (en) | A kind of people counting method based on improved LBP operator | |
Mahantesh et al. | An impact of complex hybrid color space in image segmentation | |
CN106603888A (en) | Image color extraction processing structure | |
Ahmad et al. | SSH: Salient structures histogram for content based image retrieval | |
CN109544614B (en) | Method for identifying matched image pair based on image low-frequency information similarity | |
Obulesu et al. | Image retrieval based local motif patterns code | |
Kothyari et al. | Content based image retrieval using statistical feature and shape extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140730 |
|
WD01 | Invention patent application deemed withdrawn after publication |