基于NSCT和形态学的微透镜表面分割

Segmentation of Microlens Surfaces Based on Contourlet Transform and Mathematical Morphology

摘要:

为有效分割精密光学微透镜及微透镜阵列的表面不同特征区域,结合具有良好平移不变性、多尺度多方向分解能力的非下采样轮廓波变换(NSCT)和数学形态学运算,提出了一种微透镜表面特征边缘提取方法。采用非下采样轮廓波变换分解微透镜表面,得到粗尺度系数和精细尺度系数。粗尺度系数采用数学形态学处理,获得近似特征边缘;精细尺度系数根据变换域系数模值、硬阈值处理后得到细节边缘。融合粗尺度和精细尺度边缘作为微透镜表面的边缘。实验表明,该方法可准确识别透镜表面的特征边缘,比传统的边缘算子更为有效。

Abstract:

To segment the different feature regions on the surfaces of micro-lens array, a method is proposed for the extraction of edge features of micro-lens in combination with the mathematical morphology and the nonsubsampled contourlet transform(NSCT). Firstly, the coefficients of the coarse scales and the finer scales can be obtained after the surfaces of micro-lens are resolved using the NSCT. The approximate edges are acquired through the coarse scale coefficients which are dealt with mathematical morphology, and the detailed edges are acquired through the finer scale coefficients which are dealt with hard threshold. Thereby the edges of the coarse scales and the finer scales are merged as the edges of micro-lens. It is discovered that this method is more effective than other traditional edge detection operator in the recognition of feature edges of the lens surface.