Design and development of vision algorithms for microcrack detection in microchips

Abstract

A crack detection method for microchips is presented in this thesis. Single-channel monochrome image acquisition device was used as IC image acquiring device. Normal home personal computer is used as a processing platform for crack searching in the image acquired. Matlab ver. 6.5 was used to implement the crack detection method as suggested in this thesis. The technique used to detect cracks on a chip is referred from a book named ‘Image Processing Analysis, and Machine Vision’1. The application of graph search for edge detection was published in [Martelli 72], in which Nilsson’s A-algorithm [Nillson 82] applie. The basic ideal of this technique was implemented in border detection, but I used it in cracks searching on IC in this thesis. This thesis had demonstrated the need in implementing an algorithm for searching a Starting Node. This thesis also had shown how a program was designed to search cracks on IC. Analysis on cracks searching method presented in this thesis was made to measure the efficiency of this method. This thesis assumes all conditions of the IC images used are in optimum condition. For example, the IC smoothness, image curvature, image’s lighting condition, etc. are clearly defined. Cracks can be clearly detected if the condition of the IC images is tuned to the optimum condition.