Abstract
The template matching has been used for various industrial image applications. The normalized correlation matching is known as one of general-purposed template matching method. The main advantage of the template matching is an easy operation that requires only registration of a template image. However, a template image has been set by the manual operation. The registered template image by manual operation is not always best for matching. Especially, in the positioning and the alignment of components, the image pattern in a template image and its size influence to the reliability and computational cost. This paper proposes a method to register a template image in a referenced image. An optimal template image area is selected by the genetic algorithm in the proposed method. The experimental results show that the template images selected by the proposed method recorded higher reliability than ones registered by the manual operation.
