抄録
This paper proposes a method to recognize poses of figures which have arbitrary locations, sizes and rotations for measuring poses of two-dimensional industrial components. A template figure is represented by set of edge directions and curvature ratios of selected a pair of edge pixels and vectors that extended to the base pixel from the midpoint of a pair of pixels. Moreover, in order to cut down calculation cost by using a genetic algorithm, the characteristic portion of a template figure is extracted and it registers with a template table. The experimental results show that the poses of objective components were recognized successfully even if the components were overlapped. And, the matching time of the proposed method was cut down as compared with the case where the genetic algorithm is not used.
