A method for compressing and decompression digital image data. When compressing, the method begins by choosing a reference row. The relative data between the reference row and the current row is compared to search for the differential data set. The largest matching data set of the differential data set is searched starting from the starting position of the last differential data set. According to the matching conditions, the data type of the to-be-compressed data is first verified, then the to-be-compressed data set type and its relative referencing value are verified. Finally, the relative coding process is started. This process has a higher compression ratio and decompression speed, and increases the compression speed dramatically.