A METHOD FOR ERROR FREE SHAPE RECONSTRUCTION

Shape decomposition and reconstruction are very important aspects in image processing. The present paper proposes a new scheme which is rotationally invariant and error free algorithm, for shape representation. One of the main problems with the existing algorithms is while reconstructing the original shape is they generate noise. To overcome this, the present paper adopts a scheme that is, the current skeleton subset is chosen by comparing it with all previous skeleton subset’s. Based on this a decision will be taken whether to consider the current skeleton subset or not. By this method overlapped skeleton subsets will be removed partially, which results an error free shape representation algorithm. Various Error Functions are evaluated on the reconstructed images using the proposed method, Generalized Skeleton Transform (GST), and Octagon-Generating Decomposition (OGD) Algorithms. The present method is applied on different types of images that represent different shape components. The experimental results indicate the novelty of this algorithm in eliminating the noise.