CN110334778B - Image comprehensive similarity analysis method based on description content and image content characteristics - Google Patents

Image comprehensive similarity analysis method based on description content and image content characteristics Download PDF

Info

Publication number
CN110334778B
CN110334778B CN201910639482.8A CN201910639482A CN110334778B CN 110334778 B CN110334778 B CN 110334778B CN 201910639482 A CN201910639482 A CN 201910639482A CN 110334778 B CN110334778 B CN 110334778B
Authority
CN
China
Prior art keywords
image
content
similarity
description
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.)
Active
Application number
CN201910639482.8A
Other languages
Chinese (zh)
Other versions
CN110334778A (en
Inventor
周哲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongfang Knowledge Network Digital Publishing Technology Co ltd
Original Assignee
Tongfang Knowledge Network Digital Publishing Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tongfang Knowledge Network Digital Publishing Technology Co ltd filed Critical Tongfang Knowledge Network Digital Publishing Technology Co ltd
Priority to CN201910639482.8A priority Critical patent/CN110334778B/en
Publication of CN110334778A publication Critical patent/CN110334778A/en
Application granted granted Critical
Publication of CN110334778B publication Critical patent/CN110334778B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses an image comprehensive similarity analysis method based on description content and image content characteristics, which comprises the following steps: extracting image color information, namely RGB (red, green and blue) model and image texture information in the image content characteristic information; extracting image title content and description content in the image description content characteristic information, and performing word segmentation processing; converting the RGB model into an HSV model, and obtaining HSV values of the image to obtain 24-dimensional characteristic information of the image color; calculating the gray value of the image in the image texture information; calculating cosine similarity to obtain image content similarity; obtaining the characteristics of the image title content and the image description content; calculating the cosine similarity of the title content and the description content of the two images to obtain the similarity of the description content of the images; calculating the similarity of image features; combining the image content features and the image description content features to generate composite features of the image; calculating the cosine similarity of the image composite features to obtain the similarity of the image composite features; and when the image feature similarity and the image composite feature similarity are both larger than or equal to the threshold value, judging that the images are similar.

Description

Image comprehensive similarity analysis method based on description content and image content characteristics
Technical Field
The invention relates to a method for generating image features in image similarity analysis and a method for generating text features in text similarity analysis.
Background
In the image recognition technology, the common method is based on OCR recognition technology to match text according to the context of an image, and has limitations because the description of the text has no way to accurately express the content characteristics of the image, and the requirement of similarity detection of the image cannot be met.
Because the correctness requirement of the composite published full-text similarity analysis system on the comparison result of the image similarity is extremely high, and the existing comparison method cannot meet the requirement of the comparison on the correctness of the result, the image similarity analysis method which meets the academic unfamiliar quality requirement needs to be developed so as to achieve the purposes of improving the detection efficiency and quality and reducing the cost.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a method for analyzing comprehensive similarity of images based on the characteristics of description content and image content, which can greatly improve the efficiency of comparing image similarity and reduce the cost.
The purpose of the invention is realized by the following technical scheme:
an image comprehensive similarity analysis method based on description content and image content features comprises the following steps:
a, extracting image content characteristic information and image description content characteristic information;
b, extracting image color information, namely RGB (red, green and blue) model and image texture information in the image content characteristic information; extracting image title content and description content in the image description content characteristic information, and performing word segmentation processing;
c, converting the RGB model into an HSV model to obtain HSV values of the image, and enabling the HSV values of the image to pass through a fuzzy filter to obtain 24-dimensional characteristic information of the image color;
d, calculating the gray value of the image in the image texture information through the YIQ model, and calculating the edge texture information of the image through a digital filter, wherein the edge texture information is a 6-dimensional histogram;
e, combining 24-dimensional feature information of image colors with a 6-dimensional histogram of image edge textures to generate 144-dimensional image content features;
f, comparing the image content characteristics of the two images to be compared, and calculating cosine similarity to obtain image content similarity;
g, calculating a TF-IDF value for each word in the image title content and the description content to obtain the characteristics of the image title content and the image description content;
h, calculating the cosine similarity of the title content and the description content of the two images to obtain the similarity of the description content of the images;
i, multiplying the similarity of the image content and the similarity of the image description content by a weight value respectively to obtain the image feature similarity;
j, combining the image content characteristics with the image description content characteristics to generate composite characteristics of the image;
k, calculating the cosine similarity of the composite features of the two images to be compared to obtain the similarity of the composite features of the images;
and L, when the image feature similarity and the image composite feature similarity are both larger than or equal to the threshold value, judging that the images are similar.
One or more embodiments of the present invention may have the following advantages over the prior art:
the method has the characteristics of high feature extraction speed and small occupied space of feature descriptors, and the accuracy and efficiency are greatly improved on the basis of the original method.
Drawings
FIG. 1 is a flow chart of a method for image synthesis similarity analysis based on descriptive content and image content characteristics;
FIG. 2 is a block diagram of a method for analyzing the comprehensive similarity of images based on the description content and the characteristics of the image content;
FIG. 3 is a schematic diagram of a 10-bins blur filter;
FIG. 4 is a schematic diagram of a 24-bins blur filter;
fig. 5 is a schematic diagram of an edge histogram digital filter.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
As shown in fig. 1 and fig. 2, a flow chart of an image comprehensive similarity analysis method and a structural block diagram of the analysis method based on description content and image content features are provided. The method comprises the following steps:
step 101, extracting image content characteristic information and image description content characteristic information;
102, extracting image color information, namely RGB (red, green and blue) model and image texture information in the image content characteristic information; extracting image title content and description content in the image description content characteristic information, and performing word segmentation processing;
step 103, converting the RGB model into an HSV model to obtain HSV values of the image, and obtaining 24-dimensional characteristic information of the image color by the HSV values of the image through a fuzzy filter;
104, calculating the gray value of the image in the image texture information through a YIQ model, and calculating the edge texture information of the image through a digital filter, wherein the edge texture information is a 6-dimensional histogram;
step 105, combining 24-dimensional feature information of image colors with a 6-dimensional histogram of image edge textures to generate 144-dimensional image content features;
step 106, comparing the image content characteristics of the two images to be compared, and calculating cosine similarity to obtain image content similarity;
step 107, calculating TF-IDF values of each word in the image title content and the description content to obtain the characteristics of the image title content and the image description content;
step 108, calculating cosine similarity of the title content and the description content of the two images to obtain similarity of the image description content;
step 109, multiplying the similarity of the image content and the similarity of the image description content by a weight value respectively to obtain the image feature similarity;
step 110, combining the image content features and the image description content features to generate composite features of the image;
step 111, calculating cosine similarity of two image composite features to be compared to obtain similarity of the image composite features;
and step 112, judging that the images are similar when the image feature similarity and the image composite feature similarity are both larger than or equal to the threshold value.
1. The extraction of the image content features comprises two aspects of color and texture:
(1) when image color information is extracted, the perception and the discrimination capability of a person on colors can be well expressed by using an HSV model, wherein in the HSV model, H represents hue, S represents saturation and V represents brightness; therefore, when extracting the image color information, RGB-HSV model conversion is required to be performed on the image pixels, and the conversion formula is as follows:
V＝max(R，G，B)
Figure GDA0003071245200000041
if R is max (R, G, B) and G is equal to or greater than B,
Figure GDA0003071245200000042
if R is max (R, G, B), and G < B,
Figure GDA0003071245200000043
if G ═ max (R, G, B),
Figure GDA0003071245200000044
if B is max (R, G, B),
Figure GDA0003071245200000045
and (3) the HSV value passes through a 10-bins filter (figure 3) to obtain a 10-dimensional vector containing color information, each color region output by the 10 bins is divided into 3H value regions through a 24-bins filter (figure 4), and 24-dimensional color information is obtained through calculation.
(2) When extracting image texture information, the pixel gray value is calculated through the YIQ model, and then the edge histogram texture information of the image is extracted. The YIQ color space belongs to the NTSC system and describes the attributes of gray scale value, color, and saturation of an image, respectively. By converting the color image from RGB to YIQ space, the luminance information and chrominance information in the color image can be separated, and the RGB-YIQ correspondence is as follows:
Y＝0.29R+0.587G+0.114B
I＝0.596R+0.275G+0.321B
Q＝0.212R+0.523G+0.311B
when extracting texture information from an image, the image is divided into a plurality of cells, each of which is then subdivided into 4 sub-cells of equal size, and 5 digital filters (fig. 5 (a) (b) (c) (d) (e)) are used to extract texture edge information, and the region to which the texture edge information acts is divided into five types, i.e., vertical direction, horizontal direction, 45-degree direction, 135-degree direction, and non-direction, and the average gray values of the four sub-cells in the (< i, j) th cell are represented by go < i, j), gi < i, j), g2< i, j), and g3< i, j), respectively. a } (k), a, k, aa-as (k), ae-i3s (k) and (k) respectively represent parameters when the average gray values of four sub-cells pass through the filter, the value of each sub-cell is the parameter of the filter, wherein the value range of k is an integer from 0 to 3, and represents four sub-cells in the cell. n } (I, j), nh (I, j), } a-as (I, j, nd-I3s (1, j) and (I, j) are values of each determined direction in the (I, j) th cell.
Figure GDA0003071245200000051
Figure GDA0003071245200000052
Figure GDA0003071245200000053
Figure GDA0003071245200000054
Figure GDA0003071245200000055
Finding the maximum value
mmax＝max(nv，nh，nd-45，nd-135，nnd)
Normalizing all the n values
Figure GDA0003071245200000056
By calculation, the information of the image edge in each cell can be obtained, and the texture information is extracted as a histogram.
And finally, adding 24-dimensional color information extracted by a color module into each dimension in the vector histogram, and combining the color and the texture to finally obtain the content characteristic information of the image.
Comparing the similarity of the image contents:
after the image content features are obtained through calculation in the step 1, the image content feature values are analyzed through a cosine similarity formula, and the cosine similarity of the image content is obtained. The calculation formula is as follows:
Figure GDA0003071245200000061
left side of equal sign: cos (di, dj) represents the cosine of the included angle of the two image content characteristic vectors di, dj; on the right of equal sign, in denominator
Figure GDA0003071245200000062
The square root is then summed up after each dimension representing the vector di,
Figure GDA0003071245200000063
the square root is then summed up after each dimension representing vector dj; the numerator represents the sum of the products of each dimension of the vectors di, dj.
Extracting description contents and image titles of the images, segmenting the titles of the images and the description contents of the images to calculate weight values of all words, wherein the weight values adopt TF-IDF values of the words, and the main idea of the TF-IDF theory is as follows: the importance of a word increases in proportion to the number of times it appears in a document, but at the same time decreases in inverse proportion to the frequency with which it appears in the corpus, i.e., more specific words that appear in only a few documents are weighted more heavily than words that appear in many documents. The theory is applied in agreement of comparison between image description content and image title content, and the calculation process is as follows:
vocabulary frequency:
Figure GDA0003071245200000064
wherein N represents the total vocabulary number of the text, NWIndicating the number of times the word W appears in the image description content and the image title content, the greater the value of TF, the stronger the correlation of the word W with the text.
Inverse document frequency:
Figure GDA0003071245200000065
wherein D iswRepresenting the number of documents containing a word W, D representing the total number of documents in the corpus, the greater the IDF value, the lower the relevance of the word to the document. Computing i words in a document djThe formula of the TF-IDF value is as follows
Figure GDA0003071245200000071
After calculating the TF-IDF value of the feature item, each word in the image description content and the image title content can be represented by a vector. For example, the image description content and the image title content may be expressed as:
(1.163151,0.7855668,0.7440107,0.2310491,0.2310491,0.2288602,0.2079442,0.1938585,0.1848393,0.1653357, … …). The similarity between two image description contents and the title contents can be measured by a cosine included angle of two vectors, and the larger the value is, the higher the similarity is, therefore, the similarity between the image description contents and the image title contents can also be calculated by using a cosine similarity calculation formula (1 (2)) to finally obtain the cosine similarity of the image description contents.
And (3) carrying out comprehensive similarity comparison by combining the image content similarity and the image description content similarity:
the method comprises the following steps:
and finally, carrying out comprehensive comparison analysis on the image content feature similarity and the image description feature similarity in order to improve the accuracy of the final comparison result. And setting the similarity of the image content features and the similarity of image description as a and b. Then the final formula for calculating the result of the final synthetic similarity comparison S is as follows:
S＝k1a+k2b
wherein k is1And k2Weight value, k, for image content characteristics and image description characteristics1And k is2The sum should be 1. And finally, the larger the comprehensive similarity comparison result S is, the higher the image similarity is.
The second method comprises the following steps:
combining the image content feature values (144 dimensions) of the first portion with the image description feature values (256 dimensions) of the second portion generates a new composite vector (144+256 dimensions). When the image similarity is compared, the cosine similarity between the compound vectors of the two images is calculated by using a cosine similarity calculation formula (2 in the above 1).
And integrating the similarity analysis results of the first method and the second method, setting two threshold values, and judging the similarity of the two images when the similarity analysis results of the first method and the second method are both larger than the corresponding threshold values.
Although the embodiments of the present invention have been described above, the above descriptions are only for the convenience of understanding the present invention, and are not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (3)

1. The image comprehensive similarity analysis method based on the description content and the image content features is characterized by comprising the following steps of:
a, extracting image content characteristic information and image description content characteristic information;
b, extracting image color information, namely RGB (red, green and blue) model and image texture information in the image content characteristic information; extracting image title content and description content in the image description content characteristic information, and performing word segmentation processing;
c, converting the RGB model into an HSV model to obtain HSV values of the image, and enabling the HSV values of the image to pass through a fuzzy filter to obtain 24-dimensional characteristic information of the image color;
d, calculating the gray value of the image in the image texture information through the YIQ model, and calculating the edge texture information of the image through a digital filter, wherein the edge texture information is a 6-dimensional histogram;
e, combining 24-dimensional feature information of image colors with a 6-dimensional histogram of image edge textures to generate 144-dimensional image content features;
f, comparing the image content characteristics of the two images to be compared, and calculating cosine similarity to obtain image content similarity;
g, calculating a TF-IDF value for each word in the image title content and the description content to obtain the characteristics of the image title content and the image description content;
h, calculating the cosine similarity of the title content and the description content of the two images to obtain the similarity of the description content of the images;
i, multiplying the similarity of the image content and the similarity of the image description content by a weight value respectively to obtain the image feature similarity;
j, combining the image content characteristics with the image description content characteristics to generate composite characteristics of the image;
k, calculating the cosine similarity of the composite features of the two images to be compared to obtain the similarity of the composite features of the images;
l, when the image feature similarity and the image composite feature similarity are both larger than or equal to the threshold value, judging that the images are similar;
the similarity comparison of the image contents comprises the following steps:
after the image content features are obtained through calculation, the image content feature values are analyzed through a cosine similarity formula to obtain the cosine similarity of the image content, and the calculation formula is as follows:
Figure FDA0003071245190000021
left side of equal sign: cos (di, dj) represents the cosine of the included angle of the two image content characteristic vectors di, dj; on the right of equal sign, in denominator
Figure FDA0003071245190000022
The square root is then summed up after each dimension representing the vector di,
Figure FDA0003071245190000023
the square root is then summed up after each dimension representing vector dj; the numerator represents the sum of the products of each dimension of the vectors di, dj;
extracting description contents and image titles of the images, carrying out word segmentation on the titles and the image description contents of the images to calculate weight values of all words, wherein the weight values adopt TF-IDF values of the words, and after calculating TF-IDF values of the characteristic items, each word in the image description contents and the image title contents can be represented by a vector;
the similarity between the two image description contents and the title contents is measured by using a cosine included angle of the two vectors or calculated by using a cosine similarity calculation formula, and the cosine similarity of the image description contents is obtained, wherein the larger the value is, the higher the similarity is.
2. The method of claim 1, wherein the HSV values of the image are used to obtain 24-dimensional feature information of image colors through two blurring devices, namely a 10-bins blurring filter and a 24-bins blurring filter.
3. The method as claimed in claim 1, wherein the image texture information is extracted by dividing the image into a plurality of cells, then dividing each cell into four sub-cells with equal size, extracting the image texture edge information through a plurality of digital filters, dividing the region of action into five types of vertical direction, horizontal direction, 45 degree direction, 135 degree direction and non-direction, and determining the value of each direction through the average gray value of the cell and the filter parameters, thereby obtaining the information of the image edge in each cell.
CN201910639482.8A 2019-07-16 2019-07-16 Image comprehensive similarity analysis method based on description content and image content characteristics Active CN110334778B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910639482.8A CN110334778B (en) 2019-07-16 2019-07-16 Image comprehensive similarity analysis method based on description content and image content characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910639482.8A CN110334778B (en) 2019-07-16 2019-07-16 Image comprehensive similarity analysis method based on description content and image content characteristics

Publications (2)

Publication Number Publication Date
CN110334778A CN110334778A (en) 2019-10-15
CN110334778B true CN110334778B (en) 2021-08-06

Family

ID=68145172

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910639482.8A Active CN110334778B (en) 2019-07-16 2019-07-16 Image comprehensive similarity analysis method based on description content and image content characteristics

Country Status (1)

Country Link
CN (1) CN110334778B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111259966A (en) * 2020-01-17 2020-06-09 青梧桐有限责任公司 Method and system for identifying homonymous cell with multi-feature fusion

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002318812A (en) * 2001-04-23 2002-10-31 Olympus Optical Co Ltd Similar image retrieval device, similar image retrieval method and similar image retrieval program
CN104298749A (en) * 2014-10-14 2015-01-21 杭州淘淘搜科技有限公司 Commodity retrieval method based on image visual and textual semantic integration
CN104376105A (en) * 2014-11-26 2015-02-25 北京航空航天大学 Feature fusing system and method for low-level visual features and text description information of images in social media
CN105912642A (en) * 2016-04-08 2016-08-31 世纪禾光科技发展（北京）有限公司 Product price data acquisition method and system
CN107766349A (en) * 2016-08-16 2018-03-06 阿里巴巴集团控股有限公司 A kind of method, apparatus, equipment and client for generating text

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100451649B1 (en) * 2001-03-26 2004-10-08 엘지전자 주식회사 Image search system and method
CN101996191B (en) * 2009-08-14 2013-08-07 北京大学 Method and system for searching for two-dimensional cross-media element
CN109766465A (en) * 2018-12-26 2019-05-17 中国矿业大学 A kind of picture and text fusion book recommendation method based on machine learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002318812A (en) * 2001-04-23 2002-10-31 Olympus Optical Co Ltd Similar image retrieval device, similar image retrieval method and similar image retrieval program
CN104298749A (en) * 2014-10-14 2015-01-21 杭州淘淘搜科技有限公司 Commodity retrieval method based on image visual and textual semantic integration
CN104376105A (en) * 2014-11-26 2015-02-25 北京航空航天大学 Feature fusing system and method for low-level visual features and text description information of images in social media
CN105912642A (en) * 2016-04-08 2016-08-31 世纪禾光科技发展（北京）有限公司 Product price data acquisition method and system
CN107766349A (en) * 2016-08-16 2018-03-06 阿里巴巴集团控股有限公司 A kind of method, apparatus, equipment and client for generating text

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Android手机上图像分类技术的研究;李东阳;《中国优秀硕士学位论文全文数据库 信息科技辑》;20131115(第11期);第3章 *
IRMA-Improvisation of image retrieval with Markov chain based on annotation;G Nandha Kumar et al.;《international conference on information communication and embedded systems（CICIE2014）》;20150209;第1-7页 *

Also Published As

Publication number Publication date
CN110334778A (en) 2019-10-15

Similar Documents

Publication Publication Date Title
CN111401372B (en) Method for extracting and identifying image-text information of scanned document
Nishiyama et al. Aesthetic quality classification of photographs based on color harmony
CN104376105B (en) The Fusion Features system and method for image low-level visual feature and text description information in a kind of Social Media
CN107066972B (en) Natural scene Method for text detection based on multichannel extremal region
Feng et al. An efficient detection method for rare colored capsule based on RGB and HSV color space
CN101551823A (en) Comprehensive multi-feature image retrieval method
US10866984B2 (en) Sketch-based image searching system using cell-orientation histograms and outline extraction based on medium-level features
CN109871902A (en) It is a kind of to fight the SAR small sample recognition methods for generating cascade network based on super-resolution
CN106960182A (en) A kind of pedestrian integrated based on multiple features recognition methods again
TWI411968B (en) A method for image characterization and a method for image search
CN114359323B (en) Image target area detection method based on visual attention mechanism
Saad et al. Image retrieval based on integration between YCbCr color histogram and texture feature
CN108711160B (en) Target segmentation method based on HSI (high speed input/output) enhanced model
Sudhir et al. An efficient CBIR technique with YUV color space and texture features
CN108399335A (en) A kind of malicious code visual analysis method based on local entropy
CN108805139A (en) A kind of image similarity computational methods based on frequency-domain visual significance analysis
CN110334778B (en) Image comprehensive similarity analysis method based on description content and image content characteristics
CN103049754B (en) The picture recommendation method of social networks and device
CN104091357A (en) Method for generating mosaic images on line through subject-related images
Reddy Extraction of image features for an effective CBIR system
CN103871084B (en) Indigo printing fabric pattern recognition method
CN109784168B (en) High-resolution remote sensing transmission channel inspection method and system
CN115393748A (en) Method for detecting infringement trademark based on Logo recognition
Varish et al. A content based image retrieval using color and texture features
Jyothi et al. Computational color naming for human-machine interaction

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant