Coded data transform method, transcoding method, transcoding system, and data storage media

• US 6,590,936 B1
• Filed: 04/13/2000
• Issued: 07/08/2003
• Est. Priority Date: 04/13/1999
• Status: Expired due to Fees

• Current Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Original Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Inventors: Kadono, Shinya
• Primary Examiner(s): DIEP, NHON THANH
• Application Number:
  US09/549,380
• Time in Patent Office: 1,181 Days
• Field of Search: H04/N.712, H04/N 1.102, H04/N 1.104, 375/204.02, 375/240.03, 375/240.04, 375/240.05, 375/240.12, 375/240.13, 375/240.22, 375/240.18
• US Class Current: 375/240.12
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/40 using video transcoding, i.e. partial or full decoding of a coded input stream followed by re-encoding of the decoded output stream
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

A transcoding system comprises a decoding unit for receiving coded data which has been obtained by subjecting image data of each coding unit to a coding process including a first orthogonal transform process and a first quantization process, and decoding the input coded data for each coding unit to generate decoded data; and an encoding unit for coding the decoded data to generate transcoded data. The decoding unit includes an inverse quantizer for inversely quantizing quantized coefficients obtained from the input coded data, with a first quantization step which has been used in the first quantization process. The encoding unit includes a frequency transform unit for subjecting the decoded data to a second orthogonal transform process to generate frequency-domain data; and quantization means for subjecting the frequency-domain data to a second quantization process with a second quantization step which is derived on the basis of the transcoded data and the first quantization step. Therefore, under the restriction on the target number of bits in the transcoding process, the quantization step of the second quantization process can be set so that the quantization error is minimized, considering the non-linear relationship between the bit rate in the transcoding process and the quantization distortion (quantization error). Thereby, an increase in the quantization error due to the transcoding process can be effectively suppressed under the restriction on the bit rate.

Claims

1. A coded data transform method including a decoding process for decoding first coded data which has been obtained by subjected image data, for each coding unit, to a coding process including a first orthogonal transform process and a first quantization process to generate decoded data, and a transcoding process for coding the decoded data, for each coding unit, to generate second coded data, wherein:

said decoding process comprises;

an inverse quantization process for inversely quantizing quantized coefficients obtained from the first coded data by using a first quantization step which has been used in the first quantization process; and

said transcoding process comprises;

a second orthogonal transform process for subjecting the decoded data to orthogonal transform to generate frequency-domain data; and

a second quantization process for deriving a second quantization step on the basis of the second coded data and the first quantization step, and quantizing the frequency-domain data with the second quantization step.

2. The coded data transform method of claim 1, wherein said second quantization process further comprises:

a candidate derivation process for deriving a candidate quantization step on the basis of the second coded data; and

a quantization step derivation process for deriving the second quantization step on the basis of the candidate quantization step and the first quantization step.

3. The coded data transform method of claim 2, wherein in said quantization step derivation process:

when the candidate quantization step has a value which is equal to or larger than the value of the first quantization step and smaller than twice the value of the first quantization step, a quantization step having the same value as that of the first quantization step or a quantization step having a value equal to or larger than twice the value of the first quantization step is derived as the second quantization step; and

when the candidate quantization step has a value outside the above-described range, the candidate quantization step is derived as the second quantization step.

4. The coded data transform method of claim 1, wherein:

said coding process includes a first intra-frame coding process utilizing the correlation of pixel values in a frame and a first inter-frame coding process utilizing the correlation of pixel values between frames, wherein said first intra-frame coding process includes a first intra-frame quantization process while said first inter-frame coding process includes a first inter-frame quantization process;

said transcoding process includes a second intra-frame coding process utilizing the correlation of pixel values in a frame and a second inter-frame coding process utilizing the correlation of pixel values between frames, wherein said second intra-frame coding process includes a second intra-frame quantization process while said second inter-frame coding process includes a second inter-frame quantization process;

said decoding process includes an intra-frame decoding process adapted to the intra-frame coding process and an inter-frame decoding process adapted to the inter-frame coding process, wherein said intra-frame decoding process includes an intra-frame inverse quantization process while said inter-frame decoding process includes an inter-frame inverse quantization process; and

said second quantization process includes;

an average quantization step derivation process for deriving the average of quantization steps of all coding units in the first intra-frame quantization process as a first intra-frame average quantization step, deriving the average of quantization steps of all coding units in the second intra-frame quantization process as a second intra-frame average quantization step, deriving the average of quantization steps of all coding units in the first inter-frame quantization process as a first inter-frame quantization step, and deriving the average of quantization steps of all coding units in the second inter-frame quantization process as a second inter-frame quantization step; and

a quantization step derivation process for deriving the second quantization step on the basis of these average quantization steps and the transcoded data.

5. The coded data transform method of claim 4, wherein, in said second quantization step derivation process, the second quantization step is derived so that the ratio of the second intra-frame average quantization step to the first intra-frame average quantization step is smaller than the ratio of the second inter-frame average quantization step to the first inter-frame average quantization step.

8. A data storage medium containing a data transform program for making a computer perform the coded data transform method of claim 1.

6. A transcoding system comprising a decoding unit for receiving coded data which has been obtained by subjecting image data of each coding unit to a coding process including a first orthogonal transform process and a first quantization process, and decoding the input coded data for each coding unit to generate decoded data, and an encoding unit for coding the decoded data to generate transcoded data, wherein:

said decoding unit comprises;

an inverse quantizer for inversely quantizing quantized coefficients obtained from the input coded data with a first quantization step which has been used in the first quantization process; and

said encoding unit comprises;

a frequency transform unit for subjecting the decoded data to a second orthogonal transform process to generate frequency-domain data; and

a quantization means for subjecting the frequency-domain data to a second quantization process with a second quantization step which is derived on the basis of the transcoded data and the first quantization step.

7. The transcoding system of claim 6, wherein:

said inverse quantizer performs inverse quantization corresponding to either intra-frame quantization or inter-frame quantization according to whether the coding process performed on the input coded data is either intra-frame coding or inter-frame coding, respectively; and

said quantization means comprises;

a first step derivation unit for averaging the quantization steps of all coding units in the inverse quantization process corresponding to the intra-frame quantization process on the basis of the first quantization step supplied from said inverse quantizer, thereby deriving an intra-frame average quantization step;

a second step derivation unit for averaging the quantization steps of all coding units in the inverse quantization process corresponding to the inter-frame quantization on the basis of the first quantization step supplied from the inverse quantizer, thereby deriving an inter-frame average quantization step; and

a quantization step derivation unit for deriving a second quantization step on the basis of the transcoded data, the intra-frame average quantization step, and the inter-frame average quantization step; and

said quantization means quantizes the frequency-domain data with the second quantization step.

9. A coded data transform method comprising:

decoding first coded data obtained by subjecting image data, for each coding unit, to coding which includes a first orthogonal transform and a first quantization to generate decoded data; and

transcoding by coding the decoded data, for each coding unit, to generate second coded data;

wherein said decoding comprises;

inverse quantizing quantized coefficients obtained from the first coded data by using a first quantization step which has been used in the first quantization;

wherein said transcoding comprises;

subjecting the decoded data to orthogonal transform to generate frequency-domain data; and

deriving a second quantization step on the basis of the second coded data and the first quantization step, and quantizing the frequency-domain data with the second quantization step.

10. A coded data transform method as claimed in claim 9, wherein said deriving a second quantization step further comprises:

deriving a candidate quantization step on the basis of the second coded data; and

deriving the second quantization step on the basis of the candidate quantization step and the first quantization step.

11. A coded data transform method as claimed in claim 10, wherein in said deriving the second quantization step:

when the candidate quantization step has a value which is at least equal to the value of the first quantization step and smaller than twice the value of the first quantization step, a quantization step having the same value as that of the first quantization step or a quantization step having a value at least equal to twice the value of the first quantization step is derived as the second quantization step; and

when the candidate quantization step has a value outside the range of being at least equal to the value of the first quantization step and smaller than twice the value of the first quantization step, the candidate quantization step is derived as the second quantization step.

12. A coded data transform method as claimed in claim 9, wherein:

said coding includes a first intra-frame coding utilizing correlation of pixel values in a frame and a first inter-frame coding utilizing correlation of pixel values between frames, wherein said first intra-frame coding includes a first intra-frame quantization while said first inter-frame coding includes a first inter-frame quantization;

said transcoding includes a second intra-frame coding utilizing correlation of pixel values in a frame and a second inter-frame coding utilizing correlation of pixel values between frames, wherein said second intra-frame coding includes a second intra-frame quantization while said second inter-frame coding includes a second inter-frame quantization;

said decoding includes an intra-frame decoding adapted to the intra-frame coding and an inter-frame decoding adapted to the inter-frame coding, wherein said intra-frame decoding includes an intra-frame inverse quantization while said inter-frame decoding includes an inter-frame inverse quantization; and

said second quantization comprises;

deriving the average of quantization steps of all coding units in the first intra-frame quantization as a first intra-frame average quantization step, deriving the average of quantization steps of all coding units in the second intra-frame quantization as a second intra-frame average quantization step, deriving the average of quantization steps of all coding units in the first inter-frame quantization as a first inter-frame quantization step, and deriving the average of quantization steps of all coding units in the second inter-frame quantization as a second inter-frame quantization step; and

deriving the second quantization step on the basis of these average quantization steps and the transcoded data.

13. A coded data transform method as claimed in claim 12, wherein, in said second quantization step derivation, the second quantization step is derived so that the ratio of the second intra-frame average quantization step to the first intra-frame average quantization step is smaller than the ratio of the second inter-frame average quantization step to the first inter-frame average quantization step.

16. A computer-readable recording medium comprising a data transform program operable to make a computer perform the coded data transform method as claimed in claim 9.

14. A transcoding system comprising:

a decoding unit operable to receive coded data obtained by subjecting image data of each coding unit to coding which includes a first orthogonal transform and a first quantization, and operable to decode the input coded data for each coding unit to generate decoded data; and

an encoding unit operable to code the decoded data to generate transcoded data;

wherein said decoding unit comprises;

an inverse quantizer operable to inversely quantize quantized coefficients obtained from the input coded data with a first quantization step which has been used in the first quantization;

wherein said encoded unit comprises;

a frequency transform unit operable to subject the decoded data to a second orthogonal transform to generate frequency-domain data; and

a quantization device operable to subject the frequency-domain data to a second quantization with a second quantization step which is derived on the basis of the transcoded data and the first quantization step.

15. A transcoding system as claimed in claim 14, wherein:

said inverse quantizer is operable to perform inverse quantization corresponding to either intra-frame quantization or inter-frame quantization according to whether the coding process performed on the input coded data is either intra-frame coding or inter-frame coding, respectively; and

said quantization device comprises;

a first step derivation unit operable to average the quantization steps of all coding units in the inverse quantization corresponding to the intra-frame quantization on the basis of the first quantization step supplied from said inverse quantizer, thereby deriving an intra-frame average quantization step;

a second step derivation unit operable to average the quantization steps of all coding units in the inverse quantization corresponding to the inter-frame quantization on the basis of the first quantization step supplied from said inverse quantizer, thereby deriving an inter-frame average quantization step; and

a quantization step derivation unit operable to derive a second quantization step on the basis of the transcoded data, the intra-frame average quantization step, and the inter-frame average quantization step;

wherein said quantization device is operable to quantize the frequency-domain data with the second quantization step.