基于Gabor变换域的积分直方图鞋印图像检索  

SHOEPRINT IMAGES RETRIEVAL BASED ON INTEGRAL HISTOGRAM IN GABOR TRANSFORM DOMAIN

作　　者：黎向阳[1] 吴敏华[1] 施智平[1] 

机构地区：[1]首都师范大学信息工程学院,北京100048

出　　处：《计算机应用与软件》2015年第3期215-219,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(60903141;61165009)

摘　　要：案件现场提取的鞋印图像是现代刑侦中一种十分重要的线索和证据,如何快速而准确地从鞋印图像库中检索出相同或相似的鞋印图像,关系着公安部门的破案效率。针对有残缺的鞋印图像,提出一种基于Gabor变换域积分直方图的鞋印图像比对与检索方法。对图像进行Gabor滤波后,在每个像素位置用响应最大的滤波器号替代图像的像素值,然后统计图像积分直方图。对于残缺图像与完整图像的相似度计算,先通过积分直方图计算出残缺图像在完整图像中的最相似区域位置,再提取该区域的特征,最后进行相似度度量。通过求积分直方图快速准确地求出残缺图像在完整图像中的位置,从而提取局部特征。实验表明,与直接比较图像的全局特征相比,该算法的查准率提高了4.82%。The shoeprint images extracted at criminal scenes are a kind of important clue and evidence in modern criminal investigation.And to quickly and accurately retrieve same or similar shoeprint images from shoeprint images database relates to the criminal detection efficiency of the police. To deal with mutilated shoeprint images,we propose a shoeprint images matching and retrieval algorithm which is based on integral histogram in Gabor transform domain. After Gabor filter being applied to the image,in each pixel position the index of filter with maximum response is used instead of the pixel value of image,and then the integral histograms of image are counted. For computing the similarity of mutilated image and intact image,first,the position of mutilated image in intact image in terms of the most similar region is calculated through integral histograms. Then,the features of the region are extracted. At last,the similarity measurement is made. Through solving the integral histogram,the position of mutilated image in intact image is found out quickly and accurately,consequently the local features are extracted. Experiment demonstrates that compared with global features of directly comparing the images,the precision rate of this algorithm has an increase of 4. 82%.

关 键 词：残缺图像 GABOR滤波 积分直方图 图像检索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]