摘要 |
PURPOSE: To improve visual characteristic of an image by detecting a characteristic of a frequency distribution as to information divided for each block, deciding a quantization step size and processing the information. CONSTITUTION: Video data Vi are divided into specific sub-blocks Sb by a block processing circuit 2. The sub-blocks Sb are given to a DCT circuit, in which the data in the blocks are converted into data from a DC component to a harmonic AC component and the converted data written in a frame memory 4. On the other hand, the coefficient data given to a classification circuit 5 are classified into low and high frequency AC coefficients, after they are processed such as arithmetic operation processing by an arithmetic circuit 9, a minimum value detection circuit 13, an averaging circuit 15 and a multiplier circuit 18 or the like and reference quantization step size data Qstep are decided. The reference quantization step size data Qstep are given to a quantization circuit 19 and coefficient data of the memory frame 4 are given to the quantization circuit 19. Then the coefficient data are quantized based on the reference quantization step size data Qstep in the quantization circuit 19 and the quantized data are outputted via a variable length coding circuit 20. |