基于边缘检测的显著区域提取方法  

A Saliency Extraction Method for Color Images Based on Edge Detection

作　　者：王文正[1] 邓宸伟[1] 黄芳[1] 崔东顺[1] 赵保军[1] 

机构地区：[1]北京理工大学信息与电子学院

出　　处：《电子技术（上海）》2014年第1期14-16,共3页Electronic Technology

摘　　要：多年以来,众多专家学者建立了诸多模型来模拟人的视觉选择性注意机制,其中最具影响力的当属Itti模型,但其存在着显著区域漏检测及显著区域范围是固定形状的问题。文章基于人眼对物体的轮廓形状信息的感知能力,提出了一种改进型的显著区域提取方法,在原Itti模型基础上加入轮廓特征。本方法与原Itti模型相比较,能够改善其显著性区域的提取效果并且能够较准确的实现显著区域的分割。In the last decades, many experts and scholars establish a lot of models to simulate the human visual selective attention mechanism, of which Itti model is the most famous one. However the Itti model has such problems as detection omission of salient regions and shape of the salient regions being fixed. Since the human eye has a strong perception towards contour shape information of objects, an improved salient region extraction method is presented in this paper. In the original Itti model, we add the contour features. Compared with the original Itti model this method promotes the extraction effect of salient regions and achieves the salient region segmentation more accurately.

关 键 词：显著区域 Itti模型 视觉注意 轮廓特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]