基于过采样滤波器组的多尺度边缘检测

研究基于过采样滤波器组的多尺度边缘检测算法。图像经过过采样滤波器组分解，各个尺度下的子图像大小一致，有利于不用尺度下边缘信息的融合。提出了一种自适应双门限方法，消除了局部极大模检测中的伪边缘；同时，提出一种从大尺度边缘向小尺度边缘聚焦的方法，对不同尺度下的边缘信息进行融合，并讨论了滤波器长度对边缘检测性能的影响。仿真结果验证了所提出方法的有效性。

Multi-scale Edge Detection Based on Over-sampled Filter Banks

Multi-scale edge detection based on over-sampled filter banks is studied. The over-sampled filter banks can ensure that the size of the sub-images of different scales is same, which is of benefit to edge information fusion. An adaptive dual-thresholds scheme is proposed in order to delete the false edges from the local-maximum modulus detection. And a focusing method from large-scale edges to small-scale edges is introduced to implement edge information fusion. Furthermore, the influence of filter length on edge detection is discussed. The simulation results show the validity of the proposed method.