二维直方图准分的Tsallis熵阈值分割及其快速实现  

Precise 2-D Tsallis entropy image threshold segmentation and its fast recursive realization

作　　者：张新明[1] 郑延斌[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,新乡453007

出　　处：《仪器仪表学报》2011年第8期1796-1802,共7页Chinese Journal of Scientific Instrument

基　　金：河南省重点科技攻关项目(092102210017;102102210180)资助

摘　　要：传统二维Tsallis熵阈值法主要由于对二维直方图采用近似假设等原因,导致分割结果不够准确,由此提出了基于二维直方图准分的Tsallis熵快速图像分割方法。首先,准确选择邻域模板构建二维直方图并将Tsallis熵法用于此直方图上以便提高分割性能;然后,舍弃二维直方图中关于主对角区域的概率和近似为1的假设而准确计算使阈值选取更准确;最后,结合Tsallis熵公式对二维直方图进行分析得到其特性和2个定理,利用此特性和2个定理导出新型、快速的递推算法来降低计算复杂度。实验结果表明,与传统二维Tsallis熵法相比,所提出的方法不仅分割更准确和抗噪性更强,而且占用的存储空间和运行时间都更少。Traditional two-dimensional （2-D） Tsallis entropy （TE） thresholding methods can not obtain accurate resuits mainly owing to the supposition that the sum of the probabilities of 2-D histogram main-diagonal region is approximately one. So a fast and precise 2-D TE image thresholding method is proposed in this paper. First, a 2-O his- togram is created through accurately selecting neighborhood template and the TE method is used on the 2-D histogram to improve the segmentation performance. Then, with precise calculation and discarding the above supposition, more accurate threshold can be obtained. Finally, considering the Tsallis entropy formula, the 2-D histogram is analyzed to obtain its features and two theorems, with which a new recursive approach is inferred to reduce the computation complexity. Experimental results show that the proposed method not only achieves more accurate segmentation results and more robust anti-noise capacity, but also requires much less memory space and running time, compared with tra- ditional 2-D TE thresholding methods.

关 键 词：图像分割 阈值法 TSALLIS熵 递推算法 准分法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]