Research And Implementation Of Image Retrieval Algorithm Of Minority Costume Based On Integrated Regional Matching

The minority costume is a vividly sign of various nationalities,the witness of nation's life and the history,but also the witness of nation's life and the history.The difference of costumes constitute a national cultural diversity,is a treasure of the Chinese nation culture.At present,minority costume cultural are facing a challenge of decline and disappear.So how to effectively protect,use and inheritance is particularly important.Digitalization is the most innovative and effective form of protection.This paper analyzes the color,pattern and shape of the minority costume image from the perspective of computer vision,and explores the image processing method in view of the unique and rich characteristics.It plays a certain role in promoting the development of national culture.The main contents of this paper are as follows:The image feature extraction algorithm,multi-feature fusion algorithm and feature segmentation strategy are introduced.Then,the author briefly describes the source of the minority costume image database,analyzes the characteristics of the minority image and the description of the image feature.Then an effective multi-feature fusion algorithm is put forward.In the self-built minority tributary database,the proposed algorithm of feature extraction is applied to the minority costume retrieval.More importantly,its validity is verified.Aiming at the shortcomings of traditional global feature retrieval,an effective minority costume image retrieval algorithm based on integrated regional matching is proposed.The details are:In the image segmentation step,an improved fuzzy color feature is proposed to segment the image.In the aspect of feature extraction,putting forward a new multi-feature fusion algorithm.Then for each segmentation area were extracted different features,and multi-feature effective fusion.In addition,considering the difference of regional importance,three different weighting factors are proposed to calculate the regional weights.Based on the feedback of regional weighting adjustments,the weight used more reasonable.In the process of image matching and similarity measurement,the Integrated Region Matching algorithm(IRM)is used to search.The validity of the proposed algorithm is proved by a large number of experiments.In order to facilitate the use,this article designed and implemented a national costume image retrieval system.The system is a based on the application of comprehensive matching algorithm based on the regional.It is also use the Web platform,fused technology of image segmentation,feature extraction and region matching.According to the software model,designing and completing the retrieval system.The interface of minority clothing retrieval system is simple,easily operate,stronger versatility.It is also provide a reference for other researchers.