A cross-media Hash retrieval method and system under the guidance of multiple serialized features

A cross-media, multi-feature technology, applied in the direction of unstructured text data retrieval, digital data information retrieval, text database indexing, etc., can solve the problems of underutilization, increase diversity, reduce the impact of results, and facilitate retrieval performance effect

Active Publication Date: 2019-04-23
PEKING UNIV
View PDF2 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, none of the above methods make full use of the information of different scale features in the media, such as texture, object, scene information in the image, sentence and wor

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
  • A cross-media Hash retrieval method and system under the guidance of multiple serialized features
  • A cross-media Hash retrieval method and system under the guidance of multiple serialized features
  • A cross-media Hash retrieval method and system under the guidance of multiple serialized features

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0034] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0035] The serialized multi-feature-guided cross-media hash retrieval method of the present invention, its flow is as follows figure 1 shown, including the following steps:

[0036] (1) Establish a database of images and texts, and extract features of images and texts at multiple scales.

[0037] The image feature vectors are specifically: POOL-5, FC-6, and FC-7 layer features of the VGG-19 network, which respectively represent the underlying primitive scale features, middle-level concept scale features, and high-level semantic scale features; the text features are specifically: text words The bag-of-words feature and the bag-of-words feature of each sentence are dimensionally reduced and concatenated using the principal component analysis method, representing word-scale features and sentence-scale features, respectively.

[0038] (...

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 relates to a cross-media Hash method and system based on serialized multi-feature guidance. The method comprises the following steps of 1 establishing a database of images and texts, andrespectively extracting features of the images and the texts under various scales; 2 inputting different scale characteristics of the image and the text into two paths of recurrent neural networks according to a set sequence, and calculating Hash codes of the image and the text; and 3 optimizing network parameters through the inter-scale association constraint function and the inter-media and intra-media Hash constraint functions to realize inter-scale association mining and Hash function learning, and 4 in the retrieval stage, extracting different scale characteristics of the query image ortext, and generating corresponding Hash codes according to the same method in the step 2 to realize cross-media Hash retrieval. According to the method, the association relationships among various scales of different media can be mined to realize learning of the Hash function, and the higher retrieval accuracy is achieved compared with an existing method.

Description

technical field [0001] The invention relates to cross-media hash retrieval between images and texts, in particular to a serialized multi-feature guided cross-media hash retrieval method and system. Background technique [0002] Cross-media retrieval is a highly flexible retrieval method, users can use any kind of media to retrieve related data of other media types. With the increase of Internet data, retrieval efficiency has gradually become an important requirement in practical applications. Cross-media hash retrieval refers to the mapping of multimedia data into a unified Hamming space, using a short Hamming code, which can not only greatly improve the speed of cross-media retrieval, but also greatly compress the required storage space, which has important applications value. [0003] Traditional text or image hash retrieval methods encounter the problem of "heterogeneous gap" when facing cross-media retrieval tasks, that is, data of different media types are distributed...

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): G06F16/583G06F16/31G06F17/27G06K9/62
CPCG06F40/30G06F18/22
Inventor 彭宇新叶钊达
Owner PEKING 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