METHOD OF SELECTING RE-QUANTIZATION STEP SIZE WHOSE BIT RATIO IS SUDDENLY CHANGED AND BIT RATIO CONTROL METHOD USING THE SAME

摘要

PURPOSE: A method of controlling bit ratios using a re-quantization step size is provided to use a piecewise linearly decreasing model to control the bit ratios, so as to perform an excellent bit ratio control with a much smaller calculation amount than an existing method. CONSTITUTION: An algorithm initializes a normalized bit generation amount and a bit generation amount for re-quantization(71). The algorithm decides a section including the size of the normalized bit generation amount(Rr)(72). If the Rr is smaller than 1, the algorithm finds a Q2 value for each macro block from a formula 13(73). If the Rr is smaller than beta, the algorithm finds a Q value for each macro block from a formula 15(74). If the Rr is smaller than gamma, the algorithm finds a Q2 value for each macro block from a formula 16(75). If the Rr is delta, the algorithm finds a Q2 value for each macro block from a formula 17(76). The algorithm finds the number of macro blocks to have Q2 values of a Qd1 and a Qf1(77). The algorithm finds the Qf1 or the Qd1 as Q2 values for each macro block by using a formula 1 or formula 3(78). The algorithm re-quantizes the Qf1 and the Qd1 values as Q2 values for each macro block to code the re-quantized values(79). The algorithm updates a variable for transcoding a next slice according to a formula 20(80).