Parametric Circle Approximation Using Genetic Algorithms

摘要: Pattern recognition is one of the main research areas in computer vision community. The basic problem consists detecting and recognizing or several known patterns a data stream. Patterns can be specified raw space any feature suitable for analysis input data. In particular, visual pattern deals with applications where description terms information comes from kind sensor (Chen & Wang, 2005). A common classification scheme divides according to way defined. There are then structural statistical techniques. With respect solving approach, object location techniques divided into two types methods: i) deterministic methods like Hough transform, e.g. (Yuen et al., 1990), geometric hashing template model matching, (Iivarinen 1997; Jones 1990) ii) stochastic techniques, including RANSAC (Fischer Bolles, 1981), simulated annealing genetic algorithms (Roth Levine, 1994). Geometric shapes very useful number tasks because they often present human-made environments. They also widely used as part man-designed symbols. To recognize this shapes, many have been developed. circle ellipse detection problems studied shape Most approaches use transform-based For instance, Lam Yuen (Lam Yuen, 1996) proposed hypothesis filtering approach transform detect circles images. Lo Lo, 1994) posed multi-resolution transform. traffic sign detector by Mainzer uses (Mainzer, 2002a,b). Shape using soft computing addressed Rosin Nyongesa (Rosin Nyongesa, 2000). Genetic Algorithms (GA), Holland (Holland, 1975) 60s, family we apply an artificial evolution process. purpose produce computational individual which best fitted solve specific problem. GA extensively optimization problems. (GA) already tasks. order them problem, need pose We associate key elements both approaches. Bandyopadhay