A novel sampling scheme for image compression

Abstract

The transform approach of image data compression, as a lossy technique, depends on rejection of redundant coefficients, which are distributed depending on the transform used as well as covariance structure of the processed image. The threshold sampling scheme selects the coefficients above a predefined threshold, while, zonal sampling scheme selects the coefficients inside a predefined geometric zone. In this paper, a new scheme for the zone construction of the selected coefficients is presented, in this procedure, the higher-value coefficients are selected using a mathematical formulation. The experimental results, using a group of transforms are compared with that of the other two sampling schemes employing the same transforms. The comparison demonstrates the effectiveness of this method.