Gradient descent approach for minimizing dissimilarity measure in log-polarimagery  

作　　者：JIN Yong-jun JIANG You-yi 

机构地区：[1]College of Astronautics, Northwestern Polytechnical University, Xi＇an 710072, China

出　　处：《Optoelectronics Letters》2006年第4期302-304,共3页光电子快报（英文版）

基　　金：SupportedbyNationalNaturalScienceFoundationofChina(GrantNo.60575013).

摘　　要：Log-polar mapping has been proposed as a very appropriate space-variant imaging model in active vision applications.There is no doubt about the importance of translation estimation in active visual tracking.In this paper an approach is presented,and its performances are evaluated.The approach uses a gradient descent for minimizing a dissimilarity measure.The experimental results reveal that this method is efficient for approaching active image translations.

关 键 词：图象处理 空间变量 图象模型 视频跟踪 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]