Iterative Spline approximation for high quality image data compression
Sai Yang, Kazuo Toraichi, Iwao Sekita, Shigeo Yoneda, Ryoichi Mori · 1st IASTED International Symposium on Signal Processing and its Applications · 1987
This paper presents a method of compressing digital image data under the condition that compressed data keep prescribed fidelity, and of reconstructing a image at high speed from the compressed data. Compression rate varies automatically with complexity of texture of original image and with prescribed fidelity of reconstructed image. In experiments, compression rate of a girl's face is 33% and that of a car is 79% under the condition where we hardly notice the difference between original images and reconstructed images. The present, method is classified into three stages: (1) extracting the intervals which is suitable for functional approximation, (2) approximating the extracted intervals by quadratic spline functions in least square rule, and storing B-spline coefficients as compressed image data, and (3) reconstructing an image at high speed by using the merit of equispaced knots.