A Fast B_Spline Based Algorithm for Image Zooming and Compression

Abstract

An efficient non-real time method is proposed for evaluating the output of non-causal digital systems. This method is used for the evaluation of B_spline coefficients fitting a given data. Next, it is shown that using a decimated version of the signal or image under consideration, the number of these coefficients can be appreciably reduced and their values can be optimally chosen without any appreciable degradation of the reconstructed signal/image using these reduced number of coefficients. Applications to image zooming and compression are also suggested. Image compression is based on compressing these B_spline coefficients rather than the image itself. Illustrative examples are given, to verify the high quality of the reconstructed images when compared to other compression techniques.