Contour Extraction and Compression Scheme Utilizing Both the Transform and Spatial Image Domains

Abstract

Two new simple but fast and pretty efficient approaches for contour data detection, extraction and approximation are presented in this paper. The High-Pass Filter (HPF) method, designed to detect and extract contours from greyscale images is the first presented method. It operates in spectral domains either of the Periodic Haar Piecewise-Linear (PHL) transform or the Haar Wavelet one. The other presented method, known as the Segments Distances Ratios (SDR) approach, is used, in turn, to approximate the contour lines given by the HPF method. Its spatial approximation accuracy is carefully investigated and reported as well as referred to the universally recognized Ramer algorithm. Efficiency of both presented methods as well as their performance aspects are finally discussed and concluded.