BITRATE REDUCTION TECHNIQUES FOR IMAGE TRANSCODING

• US 20080298469A1
• Filed: 10/05/2007
• Published: 12/04/2008
• Est. Priority Date: 05/31/2007
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Liu, Shizhong, Nguyen, Ricky
• Granted Patent: US 8,213,498 B2
• Time in Patent Office: Days
• US Class Current: 375/240.26
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/132 Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
• CPC Class Codes: H04N 19/15 by monitoring actual compressed data size at the memory before deciding storage at the transmission buffer
• CPC Class Codes: H04N 19/164 Feedback from the receiver or from the transmission channel
• CPC Class Codes: H04N 19/17 the unit being an image region, e.g. an object
• CPC Class Codes: H04N 19/18 the unit being a set of transform coefficients
• CPC Class Codes: H04N 19/186 the unit being a colour or a chrominance component
• 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

This disclosure describes techniques for encoding data to compress image data to a reduced size. The techniques may be particularly effective in transcoding an image to a reduced size for transmission. An encoding device may transcode an image to an acceptable file size using bitrate control techniques that affect the entire image uniformly to transcode the image. For example, the encoding device may modify one or more quantization tables as a function of the near linear relationship between bitrate (R) and the number of zero-valued quantized transform coefficients to reduce the file size uniformly over the entire image. Alternatively, or additionally, the encoding device may transcode the image using bitrate control techniques that affect only specific localized regions of the image.

Claims

1. A method of processing digital image data comprising:

scaling one or more quantization tables as a function of a computed scaling factor;

encoding a plurality of blocks of pixels of the digital image data using the one or more scaled quantization tables;

tracking a difference between a number of bits encoded at an n^th block of the plurality of blocks of pixels of the digital image data and a number of bits that ideally should be encoded at the n^th block to achieve a target bitrate; and

removing one or more bits used to encode a portion of the plurality of blocks of the image data when the difference is greater than or equal to a threshold.

2. The method of claim 1, further comprising:

identifying a desired number of zero-valued quantized transform coefficients to achieve the target bitrate based on a bitrate of the image data as initially encoded, a number of zero-valued quantized transform coefficients of the image data as initially encoded, and the target bitrate; and

computing the scaling factor for use in scaling the one or more quantization tables based on the desired number of zero-valued quantized transform coefficients.

3. The method of claim 1, wherein scaling the one or more quantization tables comprises scaling a luma quantization table and a chroma quantization table, wherein the chroma quantization table is scaled more than the luma quantization table.

4. The method of claim 3, wherein scaling the luma quantization table and the chroma quantization table comprises:

scaling a first portion of the luma quantization table by the computed scaling factor; and

scaling a second portion of the luma quantization table and the chroma quantization table by a second scaling factor, wherein the second scaling factor is determined as a function of the computed scaling factor,

wherein the computed scaling factor and second scaling factor are greater than or equal to one, and further wherein the computed scaling factor is smaller than the second scaling factor.

5. The method of claim 4, further comprising:

ordering quantization entries of the luma quantization table and the chroma quantization table into a single sequence beginning with low frequency quantization entries of the luma quantization table followed by high frequency quantization entries of the luma quantization table then low frequency quantization entries of the chroma quantization table and then high frequency quantization entries of the chroma quantization table;

scaling the first k quantization entries of the sequence by the computed scaling factor; and

scaling the remaining N−

k quantization entries of the sequence by the second scaling factor, wherein N is the total number of quantization entries of the sequence.

6. The method of claim 1, wherein removing the one or more bits comprises:

selecting one or more quantized transform coefficients that have values that are equal to or close to a power of two; and

decrementing the one or more selected quantized transform coefficients until the selected quantized transform coefficients are less than the respective power of two.

7. The method of claim 6, wherein selecting one or more quantized transform coefficients that are equal to or close to a power of two comprises selecting one or more quantized transform coefficients that will not equal zero when decremented to a value below the respective power of two.

8. The method of claim 1, wherein removing the one or more bits comprises:

identifying at least one bit sequence that includes a plurality of bits arranged in a manner to act as a marker that requires stuffing of one or more bits; and

changing at least one of the plurality of bits of the bit sequence that corresponds to a residue such that the bit sequence no longer acts as the marker.

9. The method of claim 1, further comprising encoding the blocks of the image data using the one or more scaled quantization tables without removing the one or more additional bits when the difference falls below the threshold.

10. An apparatus for processing digital image data comprising:

a memory that stores one or more quantization tables for use in encoding a plurality of blocks of pixels of the digital image data; and

an encoding module that scales the one or more quantization tables as a function of a computed scaling factor, encodes a plurality of blocks of pixels of the digital image data using the one or more scaled quantization tables, tracks a difference between a number of bits encoded at an n^th block of the plurality of blocks of pixels of the digital image data and a number of bits that ideally should be encoded at the n^th block to achieve a target bitrate, and removes one or more bits used to encode a portion of the plurality of blocks of the image data when the difference is greater than or equal to a threshold.

11. The apparatus of claim 10, wherein the encoding module orders quantization entries of a luma quantization table and a chroma quantization table into a single sequence beginning with low frequency quantization entries of the luma quantization table followed by high frequency quantization entries of the luma quantization table then low frequency quantization entries of the chroma quantization table and then high frequency quantization entries of the chroma quantization table, scales the first k quantization entries of the sequence by the computed scaling factor, and scales the remaining N−

k quantization entries of the sequence by a second scaling factor determined as a function of the computed scaling factor,

wherein N is the total number of quantization entries of the sequence and

further wherein the computed scaling factor and second scaling factor are greater than or equal to one, and the computed scaling factor being smaller than the second scaling factor.

12. The apparatus of claim 10, wherein the encoding module selects one or more quantized transform coefficients that have values that are equal to or close to a power of two and decrements the one or more selected quantized transform coefficients until the selected quantized transform coefficients are less than the respective power of two.

13. The apparatus of claim 10, wherein the encoding module identifies at least one bit sequence that include a plurality of bits arranged in a manner to act as a marker that requires stuffing of one or more bits and changes at least one of the plurality of bits of the bit sequence such that the bit sequence no longer acts as the marker.

14. The apparatus of claim 10, wherein the encoding module resumes encoding of the blocks of the image data using the scaled quantization tables without removing the one or more additional bits when the difference falls below the threshold.

15. The apparatus of claim 10, wherein the apparatus is incorporated within a wireless communication device handset, the wireless communication device handset further comprising a transmitter for transmitting the encoded the encoded blocks of pixels.

16. An apparatus for processing digital image data comprising:

means for scaling one or more quantization tables as a function of a computed scaling factor;

means for encoding a plurality of blocks of pixels of the digital image data using the one or more scaled quantization tables;

means for tracking a difference between a number of bits encoded at an n^th block of the plurality of blocks of pixels of the digital image data and a number of bits that ideally should be encoded at the n^th block to achieve a target bitrate; and

means for removing one or more bits used to encode a portion of the plurality of blocks of the image data when the difference is greater than or equal to a threshold.

17. A computer-program product for processing digital image data comprising a computer readable medium having instructions thereon, the instructions comprising:

code for scaling one or more quantization tables as a function of a computed scaling factor;

code for encoding a plurality of blocks of pixels of the digital image data using the one or more scaled quantization tables;

code for tracking a difference between a number of bits encoded at an n^th block of the plurality of blocks of pixels of the digital image data and a number of bits that ideally should be encoded at the n^th block to achieve a target bitrate; and

code for removing one or more bits used to encode a portion of the plurality of blocks of the image data when the difference is greater than or equal to a threshold.

18. The computer-program product of claim 17, wherein the instructions further compnse:

code for identifying a desired number of zero-valued quantized transform coefficients to achieve the target bitrate based on a bitrate of the image data as initially encoded, a number of zero-valued quantized transform coefficients of the image data as initially encoded, and the target bitrate; and

code for computing the scaling factor for use in scaling the one or more quantization tables based on the desired number of zero-valued quantized transform coefficients.

19. The computer-program product of claim 17, wherein code for scaling the one or more quantization tables comprises code for scaling a luma quantization table and a chroma quantization table, wherein the chroma quantization table is scaled more than the luma quantization table.

20. The computer-program product of claim 19, wherein code for scaling the luma quantization table and the chroma quantization table comprises:

code for scaling a first portion of the luma quantization table by the computed scaling factor; and

code for scaling a second portion of the luma quantization table and the chroma quantization table by a second scaling factor, wherein the second scaling factor is determined as a function of the computed scaling factor,

wherein the computed scaling factor and second scaling factor are greater than or equal to one, and further wherein the computed scaling factor is smaller than the second scaling factor.

21. The computer-program product of claim 20, wherein code for scaling the luma quantization table and the chroma quantization table comprises:

code for ordering quantization entries of the luma quantization table and the chroma quantization table into a single sequence beginning with low frequency quantization entries of the luma quantization table followed by high frequency quantization entries of the luma quantization table then low frequency quantization entries of the chroma quantization table and then high frequency quantization entries of the chroma quantization table;

code for scaling the first k quantization entries of the sequence by the computed scaling factor; and

code for scaling the remaining N−

k quantization entries of the sequence by the second scaling factor, wherein N is the total number of quantization entries of the sequence.

22. The computer-program product of claim 17, wherein code for removing the one or more bits comprises:

code for selecting one or more quantized transform coefficients that have values that are equal to or close to a power of two; and

code for decrementing the one or more selected quantized transform coefficients until the selected quantized transform coefficients are less than the respective power of two.

23. The computer-program product of claim 22, wherein code for selecting one or more quantized transform coefficients that are equal to or close to a power of two comprises code for selecting one or more quantized transform coefficients that will not equal zero when decremented to a value below the respective power of two.

24. The computer-program product of claim 17, wherein code for removing the one or more bits comprises:

code for identifying at least one bit sequence that includes a plurality of bits arranged in a manner to act as a marker that requires stuffing of one or more bits; and

code for changing at least one of the plurality of bits of the bit sequence such that the bit sequence no longer acts as the marker.

25. The computer-program product of claim 17, further comprising code for resuming the encoding of the blocks of the image data using the one or more scaled quantization tables without removing the one or more additional bits when the difference falls below the threshold.