高速高精度的机器视觉定位的算法

New algorithm for fast and precise localization of machine vision

针对IC装备的视觉系统在复杂条件下很难快速、精确定位这一问题,提出了基于小波变换和径向基函数(RBF)网络算法,此算法提取图像归一化小波系数作为特征矢量,对目标图像特征进行学习,得到RBF分类器,利用它在背景图像上进行目标定位。仿真实验表明,此算法可以实现在旋转、平移、尺度变化和噪声条件下的快速定位,定位精度达到亚像素级。

A new algorithm, based on wavelet transform and radial basis function （RBF） network algorithm is proposed to resolve the localization problem of a machine vision system. The normalized coefficients of wavelet features are extracted by image transformation. RBF searches the process based on RBF principle and implements rough, fast and accurate pattern localization in the background. Experimental results show that it is non-sensitive to translation, rotation, scaling and noises. The localization error is less than 0.1 pixel.