DATA COMPRESSION USING BLOCK LIST TRANSFORM

摘要

Input data point signals belonging to data blocks are stored in a data point memory. A list of data points in each data block and a corresponding set of prediction data point designations are stored in a block list memory. An adder operates with an address counter and the block memory for addressing the data point memory. A further adder (171) and a quantizer are arranged for responding to signals output from the data point memory and from a predictor to produce quantized prediction error signals. Coded valves are developed by a combining circuit which receives the quantized prediction error signals and the predictor output signals (PRED). The coded valves are stored in the data point memory. The quantized prediction error values are passed to a variable word-length coder for non-zero values and to a zero counter for non-length coding of zero values.