Image vector quantization using geometric transform and lossless index coding

We propose an efficient image-coding scheme based on vector quantization (VQ). Typically, image qualities of the compressed images by using VQ are restricted by the codebook used. To provide better image qualities of the compressed images while keeping low bit rates, encoding the geometric transformed image block is associated with VQ. To cut down the bit rates, the index table is further processed by the lossless index coding process. According to the results, it is shown that the proposed scheme achieves better image qualities than VQ while keeping low bit rates. 1 Introduction The vector quantization (VQ) scheme 1–3 was introduced by Linde, Buzo, and Gray for grayscale image compression in 1980. Basically, VQ has a very simple decoding procedure. It is quite suitable for low-computational power devices. In addition, VQ has been applied to speech and videos coding, image hiding, image authentication, and image indexing and retrieval. VQ consists of three main procedures: codebook design, image encoding, and image decoding. The goal of the code-book design procedure is to generate a set of representative codewords. The set of representative codewords is often called the codebook. The codebook should be generated before the image encoding/decoding procedures are executed. It is well known that the Linde-Buzo-Gray (LBG) algorithm