Image compression based on Centipede Model

Abstract

An efficient contour based image coding scheme based on Centipede Model [1] is introduced in this paper. Unlike previous contour based models which presents discontinuities with various scales as a step edge of constant scale, the centipede model allows us to utilize the actual scales of discontinuities as well as location and contrast across them. The use of the actual scale of edges together with other properties enables us to reconstruct a better replica of the original image as compared to the algorithm lacking this feature. In this model, there is a centipede structure for each edge segment which lies along the edge segment and the gray level variation across an edge point is represented by a hybrid spline and distance between left and right feet of the centipede. The proposed model aims at reconstructing the closest intensity surface to the original one from the contour information. We first obtain edges by using the Generalized Edge Detector (GED) [2, 3] which controls the scale and shape of the filter, providing edges suitable to the application in hand. Then the detected edges are traced to produce the distinct contours. These contours are ranked by assigning a priority based on the weighted sum of contour length, mean contrast, contour density and mean contour curvature. In our scheme, the compression ratio is controlled by retaining the most significant segments and by adjusting the distance between the successive foot pairs. The original image is reconstructed from this sparse information by minimizing a hybrid energy functional which spans a space called lambdatau-space, where lambda represents the smoothness of the image and represents tau the continuity of the image. Since the GED filters are derived from this energy functional, we have utilized the same process both for detecting the edges and reconstructing the surface from them. The proposed model and the algorithm have been tested on both real and synthetic images. Compression ratio reaches to 180:1 for synthetic images while it ranges from 25:1 to 100:1 for real images. We have experimentally shown that the proposed model preserves perceptually important features even at the high compression ratios.