Subband seismic data compression: optimization and evaluation

A variable-length subband coder for seismic stack sections is presented. A parallel FIR filter bank is optimized to maximize the coding gain, while securing perfect reconstruction in the absence of quantization, assuming that the correlation properties of stack sections can be modelled by a separable autoregressive process. The entropy coding scheme, based on uniform quantization and dynamic arithmetic coder allocation, is optimized to maximize the rate distortion performance using a memoryless infinite Gaussian mixture distribution model for the subband signals. Comparison to a commercial wavelet based seismic coder shows that the proposed coder performs 30-70% better in terms of bit rate at a given signal-to-noise ratio. A paired comparison test verifies the superiority of the proposed subband coder, showing that the coder provides transparent visual quality at a compression ratio of 90:1