Embedded Image Coding Using Wavelet Difference Reduction

摘要: We present an embedded image coding method, which basically consists of three steps, Discrete Wavelet Transform, Differential Coding, and Binary Reduction. Both J. Shapiro’s zerotree wavelet algorithm, A. Said W. Pearlman’s codetree algorithm use spatial orientation tree structures to implicitly locate the significant transform coefficients. Here a direct approach find positions these coefficients is presented. The encoding can be stopped at any point, allows target rate or distortion metric met exactly. bits in bit stream are generated order importance, yielding fully code successively approximate original source; thus it’s well suited for progressive transmission. decoder also terminate decoding produce lower (bit) reconstruction image. Our very simple its form (which will make fast), requires no training kind prior knowledge sources, has clear geometric structure. results it quite competitive with almost all previous reported compression algorithms on standard test images.