Applications of automated defects recognition in IC packaging based on ultrasound C-SAM images

Abstract

This thesis describes the development of an automatic defects' recognition system based on image segmentation and wavelet template. Following the literature review and description of the theory, the "Susan" operator and Variational method have been tested to obtain the edge of images and are compared with the results obtained with classical operators in normal and speckle noise conditions. Then two ways to obtain the size and location of defects are obtained by using blob-coloring to process the map obtained by the Variational method.