Variable length encoding method and variable length decoding method
View Patent ↗A variable length coding method is comprised of: a coefficient value scanning step in which an RL sequence generation unit 203 , a reordering unit 202 , and a binarization unit 203 scan coefficient values within a block in a predetermined scanning order starting at a higher-frequency component toward a lower-frequency component; and an arithmetic coding step in which an arithmetic coding unit 205 and a table storage unit 204 perform arithmetic coding on the absolute values of the coefficient values according to the scanning order used in the coefficient value scanning step, by switching between probability tables 1˜4 for use, wherein, in the arithmetic coding step, a probability table to be used is switched to another probability table in one direction, when the arithmetic-coded absolute values of the coefficient values include an absolute value exceeding a predetermined threshold value.
1. A decoding method for arithmetic decoding a bit stream which is obtained by arithmetic coding, on a block basis, coefficients of a two-dimensional array of frequency components in a predetermined scanning order starting at a high frequency component toward a low frequency component, the coefficients being generated by frequency transformation performed on picture data of a block which has a predetermined size of pixels, the decoding method comprising:
an arithmetic decoding step of arithmetic decoding the bit stream into a first bit of binary data corresponding to each absolute value of the coefficients, on a block basis, according to a predetermined scanning order starting at a high frequency component toward a low frequency component by using a plurality of probability tables; and
a switching step of switching between the plurality of probability tables, from a current probability table for the first bit of the binary data corresponding to a first coefficient to be decoded, to a new probability table for the first bit of the binary data corresponding to a second coefficient to be decoded, based on a result of a comparison between an absolute value of the first coefficient to be decoded and a predetermined threshold value,
wherein, in the switching step, the switching between the plurality of probability tables is performed in a predetermined one direction within each block such that each of the probability tables, which has been used for performing arithmetic decoding on the first bit of the binary data corresponding to an already decoded coefficient before switching to the new probability table, is not used within each block after switching to the new probability table, and the switching is not performed in the direction opposite to the predetermined one direction regardless of said result of the comparison, and
wherein, within each block, if a predetermined one of the plurality of probability tables has been used to perform arithmetic decoding, the switching between the plurality of probability tables is not performed regardless of said result of the comparison.
2. The decoding method according to claim 1 ,
wherein, in the switching step, the switching to the new probability table is performed after a number of decoded coefficients having an absolute value exceeding 1 becomes a non-zero value, and the switching between the plurality of probability tables is not performed after the switching to the new probability table.
3. The decoding method according to claim 1 ,
wherein, in the arithmetic decoding step, the arithmetic decoding is performed separately on the first bit and another bit of the binary data, and a probability table used in the arithmetic decoding of said first bit of the binary data and a probability table used in the arithmetic decoding of said another bit of the binary data are different when the arithmetic decoding is performed on the binary data.
4. The decoding method according to claim 1 ,
wherein, in the switching step, the switching between the plurality of probability tables is performed in the predetermined one direction such that when a last one of the plurality of probability tables which can be used to perform arithmetic decoding is used, said last one of the probability tables is used to perform arithmetic decoding on the first bit of the binary data corresponding to each absolute value of all subsequent ones of the coefficients, and
wherein said last one of the plurality of probability tables is said predetermined one of the probability tables.
5. A computer readable recording medium having stored thereon a computer program including instructions for causing a computer to execute a decoding method for arithmetic decoding a bit stream which is obtained by arithmetic coding, on a block basis, coefficients of a two-dimensional array of frequency components in a predetermined scanning order starting at a high frequency component toward a low frequency component, the coefficients being generated by frequency transformation performed on picture data of a block which has a predetermined size of pixels,
wherein the decoding method comprises:
an arithmetic decoding step of arithmetic decoding the bit stream into a first bit of binary data corresponding to each absolute value of the coefficients, on a block basis, according to a predetermined scanning order starting at a high frequency component toward a low frequency component by using a plurality of probability tables; and
a switching step of switching between the plurality of probability tables, from a current probability table for the first bit of the binary data corresponding to a first coefficient to be decoded, to a new probability table for the first bit of the binary data corresponding to a second coefficient to be decoded, based on a result of a comparison between an absolute value of the first coefficient to be decoded and a predetermined threshold value,
wherein, in the switching step, the switching between the plurality of probability tables is performed in a predetermined one direction within each block such that each of the probability tables, which has been used for performing arithmetic decoding on the first bit of the binary data corresponding to an already decoded coefficient before switching to the new probability table, is not used within each block after switching to the new probability table, and the switching is not performed in the direction opposite to the predetermined one direction regardless of said result of the comparison, and
wherein, within each block, if a predetermined one of the plurality of probability tables has been used to perform arithmetic decoding, the switching between the plurality of probability tables is not performed regardless of said result of the comparison.
6. The computer readable recording medium according to claim 5 ,
wherein, in the switching step, the switching between the plurality of probability tables is performed in the predetermined one direction such that when a last one of the plurality of probability tables which can be used to perform arithmetic decoding is used, said last one of the probability tables is used to perform arithmetic decoding on the first bit of the binary data corresponding to each absolute value of all subsequent ones of the coefficients, and
wherein said last one of the plurality of probability tables is said predetermined one of the probability tables.