The Design And Implementation Of An Retrieval System Based On Color Emotional Semantic

Rapid development and widespread use of a variety of network applications and platforms, resulting in a large number of multimedia information, cross-media information retrieval technology is more and more people’s attention. In recent years, the image has become the main carrier of information, people on the growing requirements of image emotional semantic search. Between image low-level visual features and high-level image semantics "semantic gap" which restricts the development of image emotion retrieval technology. But image color feature is the most significant expression of the image of emotion. Thus there is an urgent need for image emotional semantic engineering of an in-depth study of the issue.As a cross-media knowledge retrieval system an important research content retrieval based on color emotional focus of the study is how the semantics of color applied to the image emotional semantic retrieval function. Which specific research and critical problem solving is to build relationships and emotional characteristics of the underlying image semantics.This paper focuses on the "use of color emotion semantic image retrieval engineering applications" theme, through analysis and research the relationship between the color characteristics of the underlying image and emotion, use psychology Plutchik color three-dimensional model of emotion, extract characteristics of the main color feature of the image, build images-color-semantic association between emotion. The system uses the standard image library Corel part of a landscape image and the author added the part of the network related images. While applying the open source package Lucene and LIRE for image retrieval feature information and image emotional semantic indexing to achieve a color emotional semantic image retrieval system based on, it can be used for a cross-media search engine. Specific work can be summarized as the following three aspects:(1) Image feature extractionThe system first color space conversion of the image, and then get the image main colors characteristics using color histogram algorithm, and then map the main colors to the corresponding color emotional space. It completed the image color characteristics and emotional characteristics the information extraction.(2) Establishment of color emotion semanticsBased on the study of color psychology, according to Plutchik color emotion model, and consider the color of socio-emotional characteristics and natural emotion, emotional color space that can be used to build Chinese characteristics of the project. Further, according to the relationship between the primary colors and color emotion space images, achieving image color and emotion semantics associated.(3)System construction based on Lucene and LIREKey technology analysis based on search engine architecture, combined with this paper color emotion semantics associated design, using the relevant technology of Lucene and LIRE, developed color emotion-based semantic retrieval system.This paper presents a method associated with color and emotion semantics, implemented a based color emotion semantic retrieval system. Enables users respectively use like the Google Knowledge Graph emotional keyword search, and Baidu knowledge map "map search map" two kinds of image emotional semantic search function. The system can be further applied to the cross-media knowledge of search engine relevant research and development.