Recognition Of Graphic Objects With Symmetric Convex Boundary And Registration Of Speckle Images

In this paper we put forward two new methods to detect LOS of a graphic object withboundary information--minimal symmetric superset based method(MSSBM) and maximalsymmetric shell intersection based method(MSSIBM).We also devised fast algorithm for themethods.To MSSBM,the sufficient and necessary condition are given, a sufficient condition andan empirical formula with the advance of easily computing are given too.MSSIBM with a verythin shell can be a substitute for symmetric Hough transform method. Based on PMBR, characterinformation of a graphic object with symmetric convex Boundary is extracted and then theidentification be implemented.The identification algorithm is easy to achieve the aim ofinvariance to scale change, tension (compression), rotation and inversion. Based on thecoefficient of shape difference in a class couple, a multi-feature graphic object "alternative"identification method is developed. Based on the MOCCOSD, a multi-feature graphic object"one of many" identification method is proposed. Identifications of rice and other seeds validatedthe methods.