一种定义感兴趣局部显著特征的新方法及其在遥感影像检索中的应用  

A New Approach for Interesting Local Saliency Features Definition and Its Application to Remote Sensing Imagery Retrieval

作　　者：朱先强[1] 黄金才[1] 邵振峰[2] 程光权[1] 

机构地区：[1]国防科学技术大学信息系统与管理学院,长沙市德雅路410073 [2]武汉大学测绘遥感信息工程国家重点实验室,武汉市珞喻路129号430079

出　　处：《武汉大学学报（信息科学版）》2013年第6期652-655,共4页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金资助项目(41201453;61201328);高等学校博士学科点专项科研基金资助项目(20114307120021)

摘　　要：提出一种定义感兴趣局部显著特征的新方法。首先根据注意力转移机制,采用迭代法提取注意焦点作为局部显著点,根据Canny边缘图与显著图的融合提取显著边缘点;然后提取显著点的色调、纹理及边缘点的方向直方图特征,用于表征兴趣目标的局部细节和轮廓信息。该方法强调影像中的显著对象,既提高了处理效率又符合人类视觉感知特性,提高了对不同影像内容的适应性。A novel local saliency features of interest extraction algorithm based on visual sali- ency measurement theory is provided. Firstly, the focuses of attention are extracted by an it- eration process, which is fitted to the attention shift mechanism. Then, we treat it as local saliency points, and the saliency edge points are got according to the combination of Canny edge graph and saliency map. After that we employ main hue, texture features of saliency points and edge orientation histogram features of edge points as feature vector for interesting objects representation. This algorithm emphasizes the salience areas in the scene, so that it can improve process efficiency and fit to human visual perceptual characteristic, otherwise this multi-channel feature describe method ensure the algorithm＇s adaptive to various scenes.

关 键 词：局部显著特征 注意焦点 显著图 边缘方向直方图 遥感影像检索 

分 类 号：P237.3[天文地球—摄影测量与遥感]