基于微分图象噪声估计的主动型边界检测方法  

An Active Edge Detection Method Based on Noise Estimation of Differential Image

作　　者：龚坚[1] 李立源[1] 陈维南[1] 

机构地区：[1]东南大学自动化研究所

出　　处：《南京航空航天大学学报》1994年第S1期136-142,共7页Journal of Nanjing University of Aeronautics & Astronautics

摘　　要：提出一种简单的主动型边界检测方法。它首先通过噪声估计捕捉图象中可能存在边界的邻域，然后获取邻域中有关边界的空间参数，由此选择最佳微分滤波算子对相应邻域进行滤波，以获取边界点。该方法具有很强的处理针对性，因此它运算速度快、精度高、一次即可获取单象素宽的边界。A new active edge detection approach is presented. After capturing theneighborhoods of discontinuity in the image and acquiring their spatial features based on noise estimation,adaptive optimal.difference operators are applied to the corresponding neighborhoods to extract the edges. The analySis and expertrment results show that the new method has the properties of hagh computational efficiency and hagh precision while assuring the edges gained to be one pixel wide.

关 键 词：计算机视觉 图象处理 噪声估计 前期视觉 边界检测 

分 类 号：TN91[电子电信—通信与信息系统]