Detection of geometric shapes by the combination of genetic algorithm and subpixel accuracy
Yaodong Wang, Noboru Funakubo · 1996
Detecting specific shape from image is an important problem in computer vision. A minimal subset is the smallest number of points (pixels) necessary to define an unique instance of a geometric primitive. To extract certain type of geometric primitives genetic algorithm has been studied. However in that method, it doesn't go far enough to detection accuracy, convergent speed and simultaneous detection of multiple shapes. In this paper, we proposed a new approach that improves detection accuracy and convergent speed for geometric shapes by the combination of genetic algorithm and subpixel accuracy (GA&SA). We also presented an algorithm to be able to implement simultaneous detection of multiple shapes based on standardized cost function and similarity between instances, taking advantage of genetic algorithm with "population search". In addition we have confirmed these practical usefulness through some experiments.