【摘要】 噪声污染的图象,区域内灰度快速变化的图象,其图象区域边界检测通常用基于局部灰度统计的方法检测,常用的最大似然比检测法是基于图象各区域灰度服从正态分布假设的参数统计检验法,而图象边界的秩和检测法则是一种非参数统计检验法。本文对以上两种方法检测图象边界的性能进行分析,提出了一种改善检得边界质量的修正秩和算法,用三幅图象做了边界检测的对比研究。在此基础上,根据待处理图象的情况及检测边界的质量要求,提出了图象边界的不同检测方法合理选用的建议,更多还原

【Abstract】 Usually for a noisy image or an image which the gray levelchange rapidely and greately within regions, the parametric or nonparametricstatistic are used to detect the image regron boundary, The Maximum likely-hood rate method for region boundary detection in image is a parametic statisticscheme based on the basic assume that the gray level within regions obey thenormal distribution. The Ranks-sum statistic technique for image regionboundary detection presented recently is one of the nonparametric statistictest method.In this paper, we analysed the performance of boundarydetection for the above two methods, A modified method to obtain the Ranks-scum is presented. Results of boundary detection experiment using the above twomethods for three images are presented. Finally a way is suggested on how tochoose a paramatric or nonparametric method properly to detect the boundariesin different images.更多还原