Template matching using multiple templates weighted normalised cross correlation
Ze-Hao Wong, Kassim Abdulrahman Abdullah, C. J. Wong · 2014
Template matching is an image comparison technique which played an important role in machine vision. In this research, a different approach to multi-template matching technique with more robust and intuitive similarity measure is described. In many studies, a numerous templates are presented to handle pattern variation while Normalised Cross Correlation (NCC) remains as the popular similarity measure. A new approach using the idea of generalised template with Weighted Normalised Cross Correlation (WNCC) based on the pixel standard deviations of templates is proposed. This approach is tested using real electronic component and similarity measure is compared with NCC in term of quality. The proposed method is found to be effective, robust and intuitive comparatively.