non's rate distortion theory [3], a better performance can be achieved by coding vectors instead of scalars, even Vector quantization (VQ) is a promising technique for low bit rate image compression. Recently image sequence compression though the source is memoryless. Nasrabadi et al. [4] have algorithms based on VQ have been reported in the literature. presented a review of the basic vector quantization tech-Image sequences are highly nonstationary and generally exhibit niques and many of their variations for image/image sevariations from frame to frame and from scene to scene; hence quence compression. using a fixed VQ codebook to encode the different frames/ In vector quantization (VQ), a set of representative imsequences may not always guarantee a good coding perforages (training set) is decomposed into L-dimensional vecmance. Several adaptive techniques which improve the coding tors. An iterative clustering algorithm such as the LBG performance have been reported in the literature. However, we algorithm [4] is used to generate a codebook CB ฯญ note that most adaptive techniques result in further increases อW i ; ฯญ 1, . . . , Nอ, where N is the size of the codebook.
in computational complexity and/or the bit rate. In this paper, The codebook is then made available at both the transa new frame adaptive VQ technique for image sequence compression (FAVQ) is presented. This technique exploits the inter/ mitter and the receiver. In the encoding process, the image intraframe correlations and provides frame adaptability at a to be compressed is decomposed into L-dimensional vecreduced complexity. A dynamic self organized codebook is used tors. For each input vector V i ฯญ อv i 1 , v i 2 , . . . , v iL อ, the to track the local statistics from frame to frame. Computer codebook CB is searched using a nearest neighbor rule to simulations using standard CCITT sequences demonstrate the find the closest codeword W j . Compression is achieved superior coding performance of FAVQ. In addition, FAVQ is by transmitting the label (index) i, corresponding to W i . a single pass technique which makes possible real time Reconstruction of images is implemented by using i as an implementation.