Method for adaptively compressing residual digital image data in a DPCM compression system

• US 5,708,511 A
• Filed: 10/17/1996
• Issued: 01/13/1998
• Est. Priority Date: 03/24/1995
• Status: Expired due to Term

• Current Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Original Assignee: Eastman Kodak Company
• Inventors: Gandhi, Bhavan R., Smith, Craig Michael, Couwenhoven, Douglas W., Rombola, Gregory, Sullivan, James R.
• Primary Examiner(s): Coles, Sr., Edward L.
• Assistant Examiner(s): BRINICH, STEPHEN M
• Application Number:
  US08/734,551
• Time in Patent Office: 453 Days
• Field of Search: 358/426-427, 358/261.1-261.4, 358/262.1, 358/432-433, 382/232, 382/238, 382/244
• US Class Current: 382/239
• CPC Class Codes: H04N 19/12 Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
• CPC Class Codes: H04N 19/30 using hierarchical techniques, e.g. scalability H04N19/63 takes precedence
• CPC Class Codes: H04N 19/50 using predictive coding H04N19/61 takes precedence
• CPC Class Codes: H04N 19/593 involving spatial prediction techniques

Abstract

A block adaptive differential pulse code modulation (DPCM) system includes both a lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals and a lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals. An encoder receives the encoder command signals from the lossless DPCM processor and lossy DPCM compressor and produces a compressed encoded bit stream. A switch responsive to a compression configuration signal and to the encoder command signals from the lossy compressor selectively passes the encoder command signals from the lossless processor or the lossy compressor to the encoder.

Claims

1. A block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a compressor for compressing the digital image by operating on n×

m blocks of pixel values, comprising;

a. a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

b. a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

c. an encoder for receiving encoder command signals and producing a compressed encoded bit stream;

d. a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder; and

e. a second lossy DPCM compressor employing a different lossy DPCM compression technique than the first lossy DPCM compressor for producing encoder command signals, the switch selectively passing the encoder command signals from the second lossy DPCM compressor to the encoder in response to the compression configuration mode signal.

2. A block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a compressor for compressing the digital image by operating on n×

m blocks of pixel values, comprising;

a. a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

b. a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

c. an encoder for receiving encoder command signals and producing a compressed encoded bit stream;

d. a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder; and

e. a second numerically lossless DPCM processor employing a different numerically lossless DPCM processing technique than the first numerically lossless DPCM processor for producing encoder command signals, the switch selectively passing the encoder command signals from the second numerically lossless DPCM processor to the encoder in response to the compression configuration mode signal.

3. A block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a compressor for compressing the digital image by operating on n×

m blocks of pixel values, comprising;

a. a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

b. a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

c. an encoder for receiving encoder command signals and producing a compressed encoded bit stream;

d. a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder; and

e. wherein the first lossless DPCM processor includes;

i. a predictor for receiving a pixel value and previously reconstructed pixel values and producing a predicted pixel value;

ii. a subtractor for subtracting the predicted pixel value from the pixel value to produce a difference pixel value;

iii. a symbolizer for generating a lossless symbol value corresponding to the difference pixel value;

iv. a classifier for receiving the lossless symbol value and producing corresponding prefix and code signals; and

v. a range select module for receiving a block of prefix signals and generating a range value for the block.

4. The block adaptive differential pulse code modulation (DPCM) system claimed in claim 3 wherein:

a. the encoder command signals from the lossless processor include a block of prefix signals identifying a symbol set from a plurality of symbol sets for each location in the block of pixel values, a block of code values identifying a symbol within the identified symbol set for each location within the block of pixel values, and a range select signal specifying the maximum prefix signal within the block of prefix signals;

b. the encoder command signals from the lossy processor include a quantizer select signal for a block of difference pixel values, and a block of symbols respectively representing a block of quantized difference pixels;

c. the switch setting a lossless compression flag when the encoder command signals are being passed from the lossless processor; and

d. the encoder produces the compressed encoded bit stream by,

when the lossless compression flag is set, using the range select signal to select an encoder table from a plurality of encoder tables, and using the prefix signals and the code values for selecting code bit patterns from the selected table for each location in the block of pixel values and encoding the range select signal as overhead information for each block of pixel values, and

when the lossless compression flag is not set, using the quantizer select signal to select an encoder table from a plurality of encoder tables, and using the symbols representing the quantized difference pixels for selecting code bit patterns from the selected table for each location in the block of pixel values, and encoding the quantizer select signal as overhead information.

5. A block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a compressor for compressing the digital image by operating on n×

m blocks of pixel values, comprising;

a. a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

b. a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

c. an encoder for receiving encoder command signals and producing a compressed encoded bit stream;

d. a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder; and

e. wherein the lossy compressor comprises;

i. a predictor for receiving a pixel value and previously reconstructed pixel values and producing a predicted pixel value;

ii. a subtractor for subtracting the predicted pixel value from the pixel value to produce a difference pixel value;

iii. run select means for receiving a block of pixel values, reconstructed pixel values and a quantizer select signal, and producing a DC track signal representing an offset for the block of pixel values, a track flag indicating whether the block of pixel values is to be DC tracked, and a run flag indicating whether the block of pixel values is classified as a run block;

iv. quantizer select means responsive to difference pixel values, and the run flag, for forming a corresponding block of difference pixel values and for producing the quantizer select signal for the block;

v. a subsample mask responsive to the quantizer select signal for producing a binary mask;

vi. a quantizer module responsive to the quantizer select signal, the binary mask, the DC track flag, the DC track signal and a second difference signal for producing quantized difference pixel values for each pixel within the block and corresponding symbol values;

vii. a second predictor responsive to reconstructed pixel values from a current line and a previous line of pixels to produce a second predicted pixel value;

viii. means for summing the second predicted pixel value and the corresponding quantized difference pixel value to produce the reconstructed pixel value; and

ix. means for subtracting the second predicted pixel value from the pixel value to produce the second difference pixel value.

6. The block adaptive differential pulse code modulation (DPCM) system claimed in claim 5, wherein the lossy compressor further comprises:

a. means for scaling the pixel value, the difference pixel value, the reconstructed pixel values from a previous line, by a scaling factor S that can be varied to control the bit rate of the lossy compressor, whereby the bit rate of the lossy compressor is increased with scale factors greater than one and is decreased with scale factors less than one; and

b. means for scaling the reconstructed pixel value from a current line by a scaling factor 1/S to produce an unscaled reconstructed pixel value.

7. The block adaptive differential pulse code modulation (DPCM) system claimed in claim 5, wherein the set of quantizers in the quantizer module are elliptically partitioned in quantizer selection space.

8. A block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a compressor for compressing the digital image by operating on n×

m blocks of pixel values, comprising;

a. a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

b. a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

c. an encoder for receiving encoder command signals and producing a compressed encoded bit stream;

d. a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder; and

e. wherein the lossy compressor further comprises means for detecting low activity regions in the digital image signal and for generating an encoder command signal instructing the encoder to employ run length coding for the low activity region.

9. An improved block adaptive differential pulse code modulation (DPCM) system for compressing a digital image signal by operating on n×

m blocks of pixel values of the type having a lossy compressor employing a set of quantizers, wherein the improvement comprises;

the set of quantizers being elliptically partitioned in quantizer selection space.

10. An improved block adaptive differential pulse code modulation (DPCM) system for compressing a digital image signal by operating on n×

m blocks of pixel values of the type having a lossy compressor employing a set of quantizers, wherein the improvement comprises;

run select means for detecting a low activity region in the digital image signal, bypassing the set of quantizers when such low activity region is detected and producing a run length signal for the low activity region.

11. The improved block adaptive DPCM claimed in claim 10, wherein the run select means includes:

a. predictor means responsive to previously reconstructed pixel values and a previous predicted pixel value for producing a predicted pixel value;

b. a subtractor for subtracting each predicted pixel value from each corresponding pixel value in the block to produce difference pixel values;

c. means responsive to the track flag and the block of difference pixel values for computing a reconstruction error by calculating the sum of the difference pixel values in the block of difference pixel values;

d. means for selecting an offset from a set of operator defined offsets in response to the reconstruction error and the track flag;

e. means for adding the selected offset to the difference pixel values in the block to produce offset difference pixel values and the DC track signal;

f. means for determining whether a block is classified as a run block in response to the offset difference pixel values; and

g. means for determining whether a block should be DC tracked based on a quantizer selected for a previous block and producing the track flag that indicates when the block should be DC tracked.

12. An improved block adaptive differential pulse code modulation (DPCM) system for compressing a digital image signal by operating on n×

m blocks of pixel values of the type having a lossy compressor employing means for forming a block of difference pixel values, a set of quantizers, means for selecting a quantizer from the set of quantizers for quantizing the difference pixel values within the block, and means for encoding the quantized difference signals, and means for transmitting the encoded quantized difference signals, wherein the improvement comprises;

a. subsample mask means responsive to a quantizer selection for producing a binary mask corresponding to the selected quantizer; and

b. means for employing the binary mask to selectively quantize, encode, and transmit the quantized difference pixel value when the corresponding mask value is one and when the corresponding mask value is zero, setting the quantized difference signal to a constant user defined value and transmitting nothing.

13. The block adaptive DPCM system claimed in claim 12, wherein the user defined value is zero.

14. In a block adaptive differential pulse code modulation (DPCM) system for transmitting or storing a digital image signal, a decompressor for reconstructing a digital image from a compressed encoded bit stream produced by a compressor of the type having a first numerically lossless DPCM processor responsive to blocks of pixel values for producing encoder command signals;

a first lossy DPCM compressor responsive to blocks of pixel values for producing encoder command signals;

an encoder for receiving encoder command signals and producing a compressed encoded bit stream; and

a switch responsive to a compression configuration mode signal and to the encoder command signals from the lossy compressor for selectively passing the encoder command signals from the numerically lossless processor or the lossy compressor to the encoder, comprising;

a. a decoder for receiving the compressed encoded bit stream and reproducing the encoder command signals and the compression configuration mode signal; and

b. reconstructing means responsive to the reproduced encoder command signals and the compression configuration mode signal for reconstructing the digital image, the reconstructing means including a first means for reconstructing lossless processed digital image data and a second means for reconstructing lossy compressed digital image data.

15. In a block adaptive differential pulse code modulation (DPCM) system, a lossless DPCM processor for compressing a digital image signal having pixel values, comprising:

a. a predictor for receiving a pixel value and previously reconstructed pixel values and producing a predicted pixel value;

b. a subtractor for subtracting the predicted pixel value from the pixel value to produce a difference pixel value;

c. a symbolizer responsive to the predicted pixel value and the corresponding difference pixel value for generating a lossless symbol value corresponding to the difference pixel value;

d. a classifier responsive to the lossless symbol value and producing corresponding prefix and code signals;

e. a range select module for receiving a block of prefix signals and generating a range value for the block; and

f. an encoder responsive to the prefix, code and range values for producing a compressed bit stream representing the digital image.

16. The lossless processor claimed in claim 15, wherein said encoder includes a Huffman encoder.

17. The lossy DPCM compressor claimed in claim 16, further comprising:

a. means for detecting low activity regions in the digital image signal; and

b. means for generating an encoder command signal instructing the encoder to employ run length coding for the low activity region.

18. The lossy DPCM system claimed in claim 17, wherein the set of quantizers in the quantizer module are elliptically partitioned in quantizer selection space.

19. In a block adaptive differential pulse code modulation (DPCM) system, a lossy DPCM compressor for compressing a digital image signal having pixel values, comprising:

a. a predictor for receiving a pixel value and previously reconstructed pixel values and producing a predicted pixel value;

b. a subtractor for subtracting the predicted pixel value from the pixel value to produce a difference pixel value;

c. run select means for receiving a block of pixel values, reconstructed pixel values and a quantizer select signal, and producing a DC track signal representing an offset for the block of pixel values, a track flag indicating whether the block of pixel values is to be DC tracked, and a run flag indicating whether the block of pixel values is classified as a run block;

d. quantizer select means responsive to difference pixel values, and the run flag, for forming a corresponding block of difference pixel values and for producing the quantizer select signal for the block;

e. a subsample mask responsive to the quantizer select signal for producing a binary mask;

f. a quantizer module responsive to the quantizer select signal, the binary mask, the DC track flag, the DC track signal and a second difference signal for producing quantized difference pixel values for each pixel within the block and corresponding symbol values;

g. a second predictor responsive to reconstructed pixel values from a current line and a previous line of pixels to produce a second predicted pixel value;

h. means for summing the second predicted pixel value and the corresponding quantized difference pixel value to produce the reconstructed pixel value;

i. means for subtracting the second predicted pixel value from the pixel value to produce the second difference pixel value; and

j. an encoder responsive to the quantizer select signal and the symbol value corresponding to the quantized difference pixel value for producing a compressed bit steam representing the digital image.

20. The lossy DPCM compressor claimed in claim 19, further comprising:

a. means for scaling the pixel value, the difference pixel value, and the reconstructed pixel values from a previous line, by a scaling factor S that can be varied to control the bit rate of the lossy compressor, whereby the bit rate of the lossy compressor is increased with scale factors greater than one and is decreased with scale factors less than one; and

b. means for scaling the reconstructed pixel value from a current line by a scaling factor 1/S to produce an unscaled reconstructed pixel value.

21. The lossy DPCM compressor claimed in claim 19, wherein the run select means includes:

a. predictor means responsive to previously reconstructed pixel values and a previous predicted pixel value for producing a predicted pixel value;

b. a subtractor for subtracting each predicted pixel value from each corresponding pixel value in the block to produce difference pixel values;

c. means responsive to the track flag and the block of difference pixel values for computing a reconstruction error by calculating the sum of the difference pixel values in the block of difference pixel values;

d. means for selecting an offset from a set of operator defined offsets in response to the reconstruction error and the track flag;

e. means for adding the selected offset to the difference pixel values in the block to produce offset difference pixel values and the DC track signal;

f. means for determining whether a block is classified as a run block in response to the offset difference pixel values; and

g. means for determining whether a block should be DC tracked based on a quantizer selected for a previous block and producing the track flag that indicates when the block should be DC tracked.

22. The lossy DPCM compressor claimed in claim 19, wherein the run select means includes:

a. subsample mask means responsive to a quantizer selection for producing a binary mask corresponding to the selected quantizer; and

b. means for employing the binary mask to selectively quantize, encode, and transmit the quantized difference pixel value when the corresponding mask value is one and when the corresponding mask value is zero, setting the quantized difference signal to a constant user defined value and transmitting nothing.

23. The block adaptive DPCM system claimed in claim 22, wherein the user defined value is zero.