Image-based navigation on a chip

Abstract

In semiconductor industry, where highest levels of precision and robustness are required, machine vision tools evolved to become a mainstream automation tools that guide robotic handling, assembly and inspection processes. This paper presents an algorithm for navigation on a chip that is based on localization of microscopic eye-point images using a previously acquired wafer map. It is fast enough for in-line microscopy and robust to visual changes occurring during the manufacturing process, such as contrast variations, re-scaling, rotation and partial feature obliteration. The algorithm uses geometric hashing, a highly efficient technique drawn from the object recognition field. Experimental results indicate high reliability of the algorithm.