微结构表面形貌的边缘保持滤波.pdfVIP

中国科技论文在线 Edge preserving filtering of micro-structured surfaces# * ZHANG Xiangchao, Xu Min (Shanghai Ultra-Precision Manufacturing Engineering, Center, Fudan Universtiy, 5 ShangHai 200438) Abstract: Micro optical elements are more and more widely used in precision engineering. Characterisation of the surface topography of these surfaces is very difficult due to the existence of sharp edges and complex features. The Gaussian filter will ruin the sharp features of structured surfaces. In this paper a Gaussian bilateral filter is developed. It uses two kernel functions. When implementing 10 filtering, the weights of the neighbouring data points are assigned not only based on the spatial closeness, but also according to the signal similarity. Thus at the regions of sharp edges, large weighting values are assigned only to the data points that are similar to the current data values, and the sharp features can be preserved successfully. Numerical validations are presented to compare the behaviours of conventional Gaussian filter and the bilateral filter. The results demonstrated that the 15 developed Gaussian bilateral filter shows remarkable superiority on f