Smoothed Contour Detection Method for Arbitrary Shape Object of Images

Abstract

Description of initial raster image objects by a set of contour vertices coordinates leads to increase efficiency of the retrieval, classification and recognition of graphical data in information systems. In this paper the smoothed contour detection method for objects of raster images is considered. The developed method is based on application of interpolation function for partial differential equation solutions and smoothed function for post-processing of detected contours. The accuracy of contour reproduction for raster image objects is increased on 4.87% by applying the developed method compared to the discrete method of contour detection.