基于ARM系统的图像检索算法与实现  

Design and Implementation of Image Retrieval Algorithms based on ARM Systems

作　　者：贝卓运[1] 张浩[1] 赵荣泳[1] 王胜华 

机构地区：[1]同济大学 CIMS 研究中心,上海200092 [2]上海复旭信息科技有限公司,上海200433

出　　处：《机电一体化》2007年第6期30-34,共5页Mechatronics

基　　金：上海市科委项目(04DZ15010)

摘　　要：针对用户对于ARM嵌入式消费类电子产品图像检索功能的需求,提出了基于特征直方图法和二分聚类算法的图像色彩特征提取方法,并根据嵌入式处理平台的特点对算法进行了优化,提高其运行效率和检索精度。通过将提取出的特征同已有特征库进行比对,实现图像的检索和分类。基于Freescale的iMX21嵌入式平台实现了图像检索系统原形,对算法进行了验证。In order to meet the demand of image retrieval in the field of embedded ARM consumer image products, two algorithms based on characteristic histogram and decomposing clustering are used to recognize color characteristics of digital images. These two algorithms are also optimized to improve their efficiency and precision according to the character- istics of embedded systems. Image retrieval and classification are realized by comparing these color characteristics with those in characteristic libraries. An application example based on Freescale iMX21 platform has also been designed to validate these algorithms.

关 键 词：ARM 图像检索 特征色提取 直方图 聚类 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]