非清晰区域抑制下的显著对象检测方法

基于上下文感知的显著区域检测模型（Context-Aware,CA）对于大目标和复杂背景图像中显著对象检测存在检测内容缺失和误检的问题.在CA模型的基础上,引入图像清晰度的视觉反差特性,提出非清晰区域抑制下的图像显著对象检测方法.该方法以离散度作为判断图像中是否存在清晰度差异的标准,并对存在差异的图像进行抑制.实验结果表明,非清晰区域抑制的CA方法可以在较好的解决大目标检测和复杂背景误检问题,提高了显著对象检测精度.

Saliency Detection Based on Inhibition of Blur Regions

There are problems of content missing and false detection for Context-Aware(CA) Saliency detection model when detecting image with large scale object or complex background.Based on Context-Aware model,with the introduction of visual contrast characteristic of image clarity,a saliency detection method based on inhibition of blur regions was proposed.The dispersion was made as the criterion to judge whether there exist clarity difference in an image,and the image would be inhibited if there exist the difference.The experimental result shows that,the saliency detection method based on inhibition of blur regions can better solve the problem of content missing and false detection in some special images with higher detection precision.