Image annotation method based on sparse group structure

A technology of image labeling and grouping, which is applied to instruments, computing, electrical and digital data processing, etc., can solve the problems that affect the labeling results, do not consider the relevance of labeling words, etc., and achieve the effect of accurate labeling results.

Active Publication Date: 2012-03-14
ZHEJIANG UNIV
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In the traditional image multi-labeling process, an independent regression model is generally constructed for each labeled word to p...

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Image annotation method based on sparse group structure
  • Image annotation method based on sparse group structure
  • Image annotation method based on sparse group structure

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0051] The image annotation method based on the sparse group structure includes the following steps:

[0052] 1) Feature extraction is performed on image data, including global features and local features. Full-play features include color histogram, color moment, color correlation map, and wavelet transform, and local features include SIFT and shape context;

[0053] 2) Each image is represented by a combination of extracted heterogeneous feature vectors. That is, an image i is denoted as (x i ,y i )∈R p × {0, 1} C , where x i =(x i1 ,...,x ip ) T ∈R p Represents the feature vector of the image, p represents the feature dimension, y i =(y i1 ,...,y iC ) T ∈ {0, 1} C is the corresponding labeled word vector, C represents the total number of labeled words in the data set, y ij = 1 means that the i-th image has the j-th label, otherwise, y ij =0. Assuming that the image has G class feature representation, d g Represents the dimension of the g-th class feature, g∈...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an image annotation method based on a sparse group structure, which comprises the following steps: 1) extracting the characteristics of an image data set; 2) selecting n pieces of data from each image data set to serve as a training set, taking the rest data as a test set, and enabling each annotated word to appear in the training set; 3) selecting the characteristics of the image by the sparse group structure; and 4) further optimizing an annotation result according to the relationship between image annotation words. In the image annotation method, the image characteristics are screened by fully utilizing the group property of image heterogeneity characteristics, the image annotation is optimized by relevance between the image annotation words, and an annotation result obtained with the image annotation method disclosed by the invention is more accurate than the annotation result obtained with the traditional annotation method.

Description

technical field [0001] The invention relates to an image labeling method based on a sparse group structure. The method utilizes sparse group structure for feature selection, combined with correlation learning between labeled words to annotate images. Background technique [0002] With the increasing maturity of image feature extraction technology, more and more heterogeneous features can be extracted, which can be used to describe multiple aspects of image visual features, such as global features (color, texture) and local features (SIFT, shape context , GLOH (Gradient Location and Orientation Histogram)). Although many heterogeneous features can be extracted from images, different heterogeneous features have different intrinsic expressive capabilities. That is to say, the combination of several types of heterogeneous features can fully express a certain semantics of an image, instead of all the heterogeneous features combined to express a certain semantics of an image. Th...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06F17/30
Inventor 吴飞庄越挺袁莹
Owner ZHEJIANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap