Embedded Laplacian pyramid still image coding using zerotrees
Frank Mueller, Klaus Illgner, Werner Praefcke · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1996
The most widely used still image coding method is given by the JPEG coding standard which works well on natural images with compression ratios up to 1:10. The algorithm we propose in this paper allows for progressive transmission, and works well even at higher compression ratios such as 1:50 - 1:100. The proposed method is in its main aspects close to the classical pyramid approach of Burt and Adelson. While retaining the main idea of using a Laplacian pyramid type decomposition, the new proposal differs in the filters employed for pyramid decomposition and in the bit allocation and quantization. For decomposition an improved Laplacian pyramid is used. Bit allocation and quantization is done jointly using a zerotree coding algorithm. The zerotree algorithm automatically performs bit allocation within the pyramid, hence adapting to the nonstationary nature of images. The encoder outputs an embedded bit stream. Hence, the decoder may truncate the bitstream at any point, resulting in a more or less detailed image. This feature is especially useful for browsing in image data bases, where the user may stop the transmission at any time.