Picture data compressing method

TLDR: In this paper, the authors proposed to control the data amount after compression by transforming the matrix of a two-dimensional DCT coefficients to a one-dimensional sequence and setting the coefficients of high frequency components after an arbitrary k-th term to zero.

Abstract: PURPOSE:To control the data amount after compression by transforming the matrix of a two-dimensional DCT coefficients to a one-dimensional sequence and setting the coefficients of high frequency components after an arbitrary k-th term to zero CONSTITUTION:For the obtained two-dimensional DCT coefficients, quantization is executed by dividing values obtained by multiplying a scale factor to the respective threshold values of a quantizing matrix composed of the 8X8 threshold values Next, for the DC component, difference from the quantized component in the preceding block is calculated and the bit number of the difference is Huffman encoded As to the AC component, zigzag scan is executed and the AC component is transformed to a one-dimensional sequence ap Afterwards, a prescribed value (k) is set to this sequence ap and a processing is executed so that the respective values of a sequence aq equipped with the relation of 'k<=q<=63' can be zero Then, the number of continuous zero is compressed by run length encoding, and two-dimensional Huffman encoding is executed using the number data of continuous zeros and the bit number of the significant coefficients