基于Z搜索共生矩阵的彩色图像检索算法  

Color image retrieval algorithm based on Z search co-occurrence matrix

作　　者：张红军 王凤领 林继明 ZHANG Hong-jun;WANG Feng-ling;LIN Ji-ming(College of Mathematics and Computer Science,Hezhou University,Hezhou 542899,China;College of Computer and Control,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]贺州学院数学与计算机学院,广西贺州542899 [2]桂林电子科技大学计算机与控制学院,广西桂林541004

出　　处：《计算机工程与设计》2018年第4期1136-1141,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61362006);广西自然科学基金项目(2014GXNSFAA108187)

摘　　要：针对图像检索中难以有效对色彩丰富的彩色图像检索、特征维数过高等问题,提出一种基于Z搜索的彩色图像检索算法。将图像的每个颜色通道划分为子块,在每个块中,通过Z搜索技术得到每对像素的强度关系,对每个颜色通道构建Z搜素的共生矩阵(Z-co-occurrence matrix,ZCM),从图像空域中提取特征并存储所有最接近的块;在相似测量阶段,通过ZCM的比较,查询图像与数据库图像的相似性,完成目标图像检索。实验结果表明,与当前检索算法相比,所提算法具有更高的查准率和查全率,在图像包含相同纹理或者色彩丰富的彩色图像时,具有更好的检索性能。For the problem that the image retrieval is difficult to search the color rich image and the feature dimension is too high,a color image retrieval algorithm based on Z search was proposed.Each color channel of the image was divided into N*N blocks,in each block,the intensity relation of each pair of pixels was obtained using Z search technique,a Z search matrix for each color channel(Z-co-occurrence matrix,ZCM)was constructed,and the features were extracted from the image space domain and each relation of the nearest block was stored.The similarity between the query image and the database image was compared through ZCM to retrieve images.Experimental results show that compared with the current search algorithm commonly used,the proposed algorithm has higher precision and recall,especially for the images contain same textures or rich color,it has better retrieval performance.

关 键 词：图像检索 Z搜索 共生矩阵 方向强度 像素强度 相似度量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]