%0 Journal Article
%T 基于SLIC超像素分割显著区域检测方法的研究<br>Salient region detection method based on SLIC superpixel segmentation
%A 汪成
%A 陈文兵
%J 南京邮电大学学报（自然科学版）
%D 2016
%X 近年来，图像显著性区域检测已经成为图像处理与分析的热点领域，RC方法是这一领域较为出色的算法之一，然而该方法存在预分割方法不精细、区域显著值分配误差较大等缺陷。为了解决上述缺陷以获得更好的显著图，文中提出了一种基于SLIC超像素分割的显著区域探测方法。该方法首先利用Mean Shift方法对输入图像进行平滑，接着利用SLIC超像素分割方法对平滑图像进行分割，然后通过改进RC方法并引入CA方法中的上下文信息特征对分割区域进行融合，最后通过区域显著值分配得到最终的显著图。实验结果表明，对比RC方法，文中所取得的显著图效果在边缘信息以及背景处理方面都有着明显的提高。在正确率和召回率2个指标上也明显优于RC方法。<br>In recent years,image salient region detection has been widely studied in the area of image processing and analysis RC method is one of the preferred algorithms at present.RC has such shortcomings as inaccurate pre-segmentation method and relatively large error of region saliency assignment.To solve these problems and obtain better saliency map,a method for salient region detection based on SLIC superpixel segmentation is proposed.The method can smooth the input image by a Mean Shift method,segregate the smooth image and integrate the segregated parts of images.The improved RC method is used to introduce immediate context and then assign the region saliency.Finally,the final saliency map comes into being.Experimental results show that compared with the RC method,the saliency map has better effect in edge information and background processing,thus outperforming yielding higher precision and better recall rates
%K 显著性探测 Mean Shift平滑 SLIC超像素分割 区域对比度&lt
%K br&gt
%K saliency detection Mean Shift smoothing SLIC superpixel segmentation regional contrast
%U http://nyzr.njupt.edu.cn/ch/reader/view_abstract.aspx?file_no=201601014&flag=1