| 摘要 |
A high bit rate Huffman decoder (300) is provided for compressed data bit stream. In an exemplary embodiment, JPEG and MPEG data compression is explained. An oversized input data register (313) receives (buss 200) sequential input data words to be decoded. Each new data word is appended to data already in the register by right shifting the new data word by the number of valid data bits in the register. A bitwise logical OR (315) operation is performed to load an operative data register (301). The operative data register is left-shifted (317) based upon the number of bits in a previous Huffman code word-coefficient pair. The left-most pair of the appropriate size is separated for examination. The separated bit strings are examined as representing the coefficient of the previous code word-coefficient pair and the current code word. The code word is used to access a Huffman lookup table (323). The lookup table provides the zeroes run length, coefficient size, and code word-coefficient pair length which is used for the next left shift. Header/marker and byte boundary information (buses 202, 204) are separately padded (303, 305, respectively) to construct same width words as the input data words. Simultaneous shifting in parallel with the data word examination preserves the relative locations of the header/markers with the data. <IMAGE> |
