Re-encoding image sets using frequency-domain differences

Abstract

Approaches to re-encoding image sets using frequency-domain differences, along with corresponding approaches to reversing the re-encoding of the image sets. For example, a re-encoding tool computes frequency-domain differences between (a) quantized coefficients for a block from encoded data in a first format and (b) quantized coefficients for a prediction of the block. The re-encoding tool entropy codes the frequency-domain differences, producing encoded data in a second format for the compressed images. Such re-encoding can significantly reduce storage cost for previously compressed images, compared to the storage cost for the previously compressed images in the first format, without hurting image quality. Subsequently, the previously compressed images in the first format can be recovered from the encoded data in the second format. Or, the encoded data in the second format can be decoded to reconstruct the images.

1 - 20 . (canceled) 
     
     
         21 . A computing device comprising:
 a processor;   a camera;   a memory storing code, which when executed by the processor causes the computing device to:
 decode first encoded data for a set of images; 
 determine a prediction relative to one or more reference images; 
 produce transform coefficients for the prediction; 
 quantize the transform coefficients for the prediction; 
 compute frequency-domain differences between the quantized transform coefficients for the prediction and corresponding quantized transform coefficients represented in the first encoded data; and 
 entropy code the frequency-domain differences. 
   
     
     
         22 . The computing device of  claim 21 , wherein producing the transform coefficients for the prediction comprises using a frequency transform. 
     
     
         23 . The computing device of  claim 22 , wherein decoding the first encoded data for the set of images comprises using an inverse frequency transform that is an inverse of the frequency transform. 
     
     
         24 . The computing device of  claim 21 , wherein decoding the first encoded data for the set of images comprises using a quantization parameter value for inverse quantization, and wherein quantizing the transform coefficients for the prediction comprises using the quantization parameter value. 
     
     
         25 . The computing device of  claim 21 , wherein, when executed by the processor, the code further causes the computing device to:
 select, for inclusion in second encoded data, the frequency-domain differences or the corresponding quantized transform coefficients.   
     
     
         26 . The computing device of  claim 25 , wherein selecting the frequency-domain differences or the corresponding quantized transform coefficients depends at least in part on a cost metric for the frequency-domain differences and a cost metric for the corresponding quantized transform coefficients. 
     
     
         27 . The computing device of  claim 21 , wherein second encoded data includes the entropy-coded frequency-domain differences, and wherein, when executed by the processor, the code further causes the computing device to transmit the second encoded data to a second computing device for decoding of the second encoded data at the second computing device. 
     
     
         28 . The computing device of  claim 21 , wherein, when executed by the processor, the code further causes the computing device to:
 reconstruct fourth encoded data from third encoded data, including:
 compute a second prediction; 
 produce second transform coefficients for the second prediction; 
 quantize the second transform coefficients for the second prediction; 
 entropy decode a second set of frequency-domain differences associated with the third encoded data; 
 combine the second set of frequency-domain differences and the second quantized transform coefficients for the second prediction to reconstruct second corresponding quantized transform coefficients; and 
 entropy coding the second corresponding quantized transform coefficients. 
   
     
     
         29 . The computing device of  claim 21 , wherein the computing device is one or more of the following: a mobile phone, a smartphone, a personal digital assistant, a tablet computer, and a laptop computer. 
     
     
         30 . One or more computer-readable memory devices storing computer-executable instructions for causing a computing system, when programmed thereby, to perform operations comprising:
 decoding first encoded data for a set of multiple images; and   re-encoding at least one image of the set of multiple images and producing second encoded data for the set of multiple images, including:
 determining a prediction relative to one or more reference images; 
 computing quantized coefficients for the prediction; 
 computing frequency-domain differences between the quantized coefficients for the prediction and corresponding quantized coefficients that are represented in the first encoded data; and 
 entropy coding the frequency-domain differences. 
   
     
     
         31 . The one or more computer-readable memory devices of  claim 30 , wherein the computing the quantized coefficients for the prediction includes performing a frequency transform on sample values of the prediction to produce coefficients for the prediction, and wherein an inverse frequency transform performed during the decoding is an inverse of the frequency transform. 
     
     
         32 . The one or more computer-readable memory devices of  claim 30 , wherein the decoding comprises inverse quantization that uses a quantization parameter value, and wherein the computing the quantized coefficients for the prediction further includes quantizing coefficients for the prediction using the quantization parameter value. 
     
     
         33 . The one or more computer-readable memory devices of  claim 30 , wherein the re-encoding further comprises:
 selecting, for inclusion in the second encoded data, the frequency-domain differences or the corresponding quantized coefficients.   
     
     
         34 . The one or more computer-readable memory devices of  claim 30 , wherein the determining the prediction, the computing the quantized coefficients, the computing the frequency-domain differences and the entropy coding are repeated on a block-by-block basis for multiple blocks of a given image of the multiple images, and wherein the determining the prediction includes performing spatial-domain disparity compensation for the prediction for a given block of the multiple blocks relative to the one or more reference images. 
     
     
         35 . The one or more computer-readable memory devices of  claim 34 , wherein at least one of the one or more reference images is adjusted using one or more of a deformation transformation and a photometric transformation. 
     
     
         36 . The one or more computer-readable memory devices of  claim 30 , wherein the computer-executable instructions further comprise instructions for causing the computing system, when programmed thereby, to perform operations comprising:
 reconstructing fourth encoded data from third encoded data, including:
 computing, with a spatial-domain disparity compensation module, a second prediction; 
 producing, with a frequency transformer, a second set of transform coefficients for the second prediction; 
 quantizing, with a quantizer, the transform coefficients for the second prediction; 
 entropy decoding, with an entropy decoder, the frequency-domain differences; 
 combining the frequency-domain differences and the quantized transform coefficients for the second prediction to reconstruct second corresponding quantized transform coefficients; and 
 entropy coding, with an entropy coder, the second corresponding quantized transform coefficients. 
   
     
     
         37 . A method for processing encoded data for a set of multiple images comprising receiving encoded data for the set of multiple images;
 computing a prediction relative to one or more reference images;   computing quantized coefficients for the prediction;   entropy decoding frequency-domain differences between the quantized coefficients for the prediction and corresponding quantized coefficients;   combining the frequency-domain differences and the quantized coefficients for the prediction to reconstruct the corresponding quantized coefficients.   
     
     
         38 . The method of  claim 37 , further comprising:
 inverse quantizing the corresponding quantized coefficients; and   performing an inverse frequency transform on the inverse quantized coefficients.   
     
     
         39 . The method of  claim 38 , wherein:
 the computing the quantized coefficients for the prediction includes performing a frequency transform on sample values of the prediction to produce coefficients for the prediction, and the inverse frequency transform is an inverse of the frequency transform; and   the computing the quantized coefficients for the prediction includes quantizing the coefficients for the prediction using a quantization parameter value, and the inverse quantizing uses the quantization parameter value.   
     
     
         40 . The method of  claim 37 , further comprising:
 entropy coding the corresponding quantized coefficients.