Abstract

A method of compressing digital image data and a device thereof of coding digital image data within constant rate data not deteriorating a resolution of an image without regard to a complexity of digital image data. A format of M×N pixel blocks (where M, N are natural numbers) is made from luminance and sub-sampled chrominance signals, and first to fourth edge values TY, TR-Y, TB-Y, Ai respectively from the luminance, sub-sampled signals and the blocks are detected. A quantization scaling factor is determined by normalizing the first edge value, and the number of bits corresponding to the each block is allocated according to the first to fourth edge values. The number of AC (alternating current) factor bits is allocated by subtracting the number of DC (direct current) factor bits of the each block coded by a one-dimensional Huffman coding from the number of the allocation bits.