A two-component image coding scheme based on two-dimensional interpolation and the discrete cosine transform

An image coding algorithm is developed which exploits some of the desirable features of the spatial and the transform domains for image data compression. This scheme is based on a two-component source model similar to some previous results [1 ,2] , but has been generalized to include two-dimensional interpolation and transform coding. Rather than processing successive scan lines of image intensity, this algorithm approximates the image intensity over adjacent subpictures using a two-dimensional polynomial and codes the residual intensities using a discrete cosine transform (DCT). Compression results with rates in the 0.5 to 1.0 bits per pixel range are demonstrated by reconstructed images and rate-distortion characteristics.