Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Automatic image marking method emerged with pseudo related feedback and index technology

A pseudo-correlation feedback and automatic image technology, applied in special data processing applications, instruments, electrical digital data processing, etc., can solve problems such as inability to obtain, limit the accuracy of keyword collection, lack of keyword text characteristics, etc.

Inactive Publication Date: 2009-09-02
BEIJING JIAOTONG UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0014] However, in the two existing tagging methods for fusion retrieval, simple initial retrieval is performed, which seriously limits the accuracy of related keyword sets and restricts the improvement of tagging performance.
Analyzing its internal reasons, the existing simple retrieval technology cannot obtain more relevant images; moreover, the ranking of keywords is also one-time, and it lacks a good use of the text characteristics of keywords themselves

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
  • Automatic image marking method emerged with pseudo related feedback and index technology
  • Automatic image marking method emerged with pseudo related feedback and index technology
  • Automatic image marking method emerged with pseudo related feedback and index technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0040] Relevant technical content and detailed description of the present invention, now cooperate accompanying drawing to explain as follows:

[0041] figure 1 It is a frame model diagram of the automatic image labeling method of the present invention which integrates pseudo-relevance feedback and retrieval. As shown in the figure, step 1: set the unlabeled image I q To query an image, search in the labeled image library to find the k nearest neighbor images, which constitute the set of related images Ψ q and the keyword set Ψ of related images w . Assuming that the top few images are related images, use I r express. Related ImagesI r with the query image I q The similarity measure p(I r |I q ) is given by:

[0042] p(I r |I q ) = ω r ·S(I r |I q ) (1)

[0043] (a)S(I r |I q ) is the relevant image I r with the query image I q The similarity between them is defined as follows:

[0044] S(I r |I q )=exp(-D(I r , I q )) ...

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 present invention provides an automatic image tagging method that integrates pseudo-relevance feedback and retrieval technology, including: Step 1: Retrieve unlabeled query images in the tagged image database, and obtain k related images and keywords of related images Set; step 2: extract the mean vector according to k related images as a new query vector, step 3: calculate each keyword, and label the posterior probability of the query image; repeat steps 1 to 3 until the preset maximum number of iterations N; Step 4: Calculate each keyword, the stability factor in each round of keyword sets; Step 5: Calculate each keyword according to the value of the posterior probability and the stability factor, and mark the ranking probability of the query image; Step 6: sort according to the obtained ranking probability of each keyword, and select the final keyword list. The invention has the advantages of improving retrieval performance and labeling accuracy, greatly improving labeling scalability, and being a flexible, reliable and practically valuable labeling method.

Description

technical field [0001] The invention relates to a novel automatic image labeling method, in particular to an automatic image labeling method which combines pseudo-correlation feedback and retrieval technology. Background technique [0002] With the emergence of a large number of multimedia, its effective management and retrieval has become an important research topic. In the 1970s and 1980s, image retrieval was mainly based on text retrieval, and the text related to images was manually annotated, which required a large workload and depended on the individual subjective judgment of the annotator. In 1992, the content-based image retrieval (CBIR-Content Based Image Retrieval) technology was first proposed, which can automatically extract the underlying features of images and automatically retrieve them. There are many CBIR systems that have been developed. However, due to the existence of the "semantic gap", the accuracy of the retrieval results is not high, which cannot mee...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06F17/30
Inventor 赵耀赵玉凤朱振峰
Owner BEIJING JIAOTONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products