Constrained-storage vector quantization with a universal codebook

S. Ramakrishnan, K. Rose, A. Gersho
1998 IEEE Transactions on Image Processing  
Many image compression techniques require the quantization of multiple vector sources with significantly different distributions. With vector quantization (VQ), these sources are optimally quantized using separate codebooks, which may collectively require an enormous memory space. Since storage is limited in most applications, a convenient way to gracefully trade between performance and storage is needed. Earlier work addressed this problem by clustering the multiple sources into a small number
more » ... into a small number of source groups, where each group shares a codebook. We propose a new solution based on a size-limited universal codebook that can be viewed as the union of overlapping source codebooks. This framework allows each source codebook to consist of any desired subset of the universal codevectors and provides greater design flexibility which improves the storageconstrained performance. A key feature of this approach is that no two sources need be encoded at the same rate. An additional advantage of the proposed method is its close relation to universal, adaptive, finite-state and classified quantization. Necessary conditions for optimality of the universal codebook and the extracted source codebooks are derived. An iterative design algorithm is introduced to obtain a solution satisfying these conditions. Possible applications of the proposed technique are enumerated, and its effectiveness is illustrated for coding of images using finite-state vector quantization, multistage vector quantization, and tree-structured vector quantization. Index Terms-Adaptive quantization, constrained storage, universal codebook, universal source coding, vector quantization. I. INTRODUCTION A. Motivation and Applications V ECTOR quantization (VQ) is an appealing coding technique because the rate-distortion bound can be approached by increasing vector dimension [1] . In applications where high reproduction quality is required, relatively high bit rates are needed. This can lead to very high complexity for such applications, because the computational and storage complexity of unstructured vector quantization grows exponentially with the rate and dimension. In applications where signals from multiple sources are to be encoded, each would Manuscript
doi:10.1109/83.679412 pmid:18276292 fatcat:fjt7j3onjzc3liis6knpuszptq