Image compression and decompression using controlled quality loss

Abstract

The loss of image quality during compression is controlled using a sequence of quality control metrics. The sequence of quality control metrics is selected for quantizing transform coefficients within an area of the image based on an error level definition. Candidate bit costs are then determined by quantizing the transform coefficients according to the error level definition or a modified error level and the sequence of quality control metrics. Where the candidate bit cost resulting from using the modified error level is lower than the candidate bit cost resulting from using the error level definition, the transform coefficients are quantized according to the modified error level and the sequence of quality control metrics. Otherwise, the transform coefficients are quantized based on the error level definition and according to the sequence of quality control metrics.

What is claimed is: 
     
         1 . A method for compressing an image, the method comprising:
 receiving an error level definition representing a maximum quantization error for compressing the image;   identifying transform coefficients to quantize within an area of the image;   selecting, based on the error level definition, a sequence of quality control metrics for controlling a quantization of the transform coefficients;   determining a first candidate bit cost by quantizing a first copy of the transform coefficients according to the error level definition and the sequence of quality control metrics;   determining a second candidate bit cost by quantizing a second copy of the transform coefficients according to a modified error level and the sequence of quality control metrics;   determining whether the second candidate bit cost is lower than the first candidate bit cost;   responsive to determining that the second candidate bit cost is lower than the first candidate bit cost, iteratively processing the transform coefficients according to the modified error level and the sequence of quality control metrics to produce quantized transform coefficients; and   producing a compressed image based on the quantized transform coefficients.   
     
     
         2 . The method of  claim 1 , wherein iteratively processing the transform coefficients according to the modified error level and the sequence of quality control metrics to produce the quantized transform coefficients comprises:
 quantizing the transform coefficients using a first quality control metric of the sequence of quality control metrics, wherein quantizing the transform coefficients using the first quality control metric results in a quantization error;   determining that the quantization error exceeds a quality score calibration associated with the first quality control metric; and   responsive to determining that the quantization error exceeds the quality score calibration, further quantizing the transform coefficients using a second quality control metric of the sequence of quality control metrics.   
     
     
         3 . The method of  claim 1 , wherein the area of the image is a first area and the sequence of quality control metrics is a first sequence of quality control metrics, the method further comprising:
 iteratively processing transform coefficients of a second area of the image using a second sequence of quality control metrics.   
     
     
         4 . The method of  claim 3 , wherein the modified error level is a first modified error level, wherein iteratively processing the transform coefficients of the second area of the image using the second sequence of quality control metrics comprises:
 determining a candidate bit cost by quantizing the transform coefficients of the second area of the image according to a second modified error level and the second sequence of quality control metrics.   
     
     
         5 . The method of  claim 1 , wherein selecting the sequence of quality control metrics for controlling the quantization of the transform coefficients comprises:
 querying a lookup table based on the error level definition.   
     
     
         6 . The method of  claim 5 , wherein the lookup table includes records associated with multiple quality control metrics, wherein selecting the sequence of quality control metrics for controlling the quantization of the transform coefficients comprises:
 determining that a subset of the multiple quality control metrics is associated with error levels exceeding the error level definition; and   excluding one or more quality control metrics of the subset of the multiple quality control metrics from the sequence of quality control metrics.   
     
     
         7 . The method of  claim 1 , wherein the error level definition is based on one of a default image compression configuration or input received from a user causing the compression of the image. 
     
     
         8 . The method of  claim 1 , wherein a quantization error represented by the modified error level is larger than the maximum quantization error. 
     
     
         9 . A method for compressing an image, the method comprising:
 selecting a sequence of quality control metrics for controlling a quantization of transform coefficients of the image;   determining a first candidate bit cost by quantizing a first copy of the transform coefficients according to the sequence of quality control metrics and a first quantization error value;   determining a second candidate bit cost by quantizing a second copy of the transform coefficients according to the sequence of quality control metrics and a second quantization error value;   based on the second candidate bit cost being lower than the first candidate bit cost, iteratively processing the transform coefficients according to the sequence of quality control metrics and the second quantization error value to produce quantized transform coefficients; and   producing a compressed image based on the quantized transform coefficients.   
     
     
         10 . The method of  claim 9 , wherein the first quantization error value corresponds to an error level definition representing a maximum quantization error for compressing the image and the second quantization error value corresponds to a modified error level different from the maximum quantization error. 
     
     
         11 . The method of  claim 10 , wherein the sequence of quality control metrics excludes one or more quality control metrics with error levels exceeding the error level definition. 
     
     
         12 . The method of  claim 9 , wherein iteratively processing the transform coefficients according to the sequence of quality control metrics and the second quantization error value to produce the quantized transform coefficients comprises:
 iterating through quality control metrics of the sequence of quality control metrics according to quality score calibrations associated with the quality control metrics.   
     
     
         13 . The method of  claim 12 , wherein iterating through quality control metrics of the sequence of quality control metrics according to quality score calibrations associated with the quality control metrics comprises:
 determining that a quantization error resulting from quantizing the transform coefficients using a first quality control metric of the quality control metrics exceeds a quality score calibration associated with the first quality control metric.   
     
     
         14 . The method of  claim 9 , wherein the transform coefficients correspond to a first region within the image, and wherein a second sequence of quality control metrics is used to produce second quantized transform coefficients based on second transform coefficients of a second region within the image. 
     
     
         15 . The method of  claim 9 , wherein the sequence of quality control metrics is selected using psychovisual modeling. 
     
     
         16 . A method for compressing an image, the method comprising:
 determining a first candidate bit cost for transform coefficients of the image according to a sequence of quality control metrics and a first quantization error value;   determining a second candidate bit cost for the transform coefficients according to the sequence of quality control metrics and a second quantization error value;   determining whether the second candidate bit cost is lower than the first candidate bit cost; and   responsive to determining that the second candidate bit cost is lower than the first candidate bit cost, iteratively processing the transform coefficients according to the sequence of quality control metrics and the second quantization error value to produce quantized transform coefficients.   
     
     
         17 . The method of  claim 16 , comprising:
 selecting the sequence of quality control metrics based on an error level definition representing the first quantization error value.   
     
     
         18 . The method of  claim 17 , wherein selecting the sequence of quality control metrics based on the error level definition representing the first quantization error value comprises:
 selecting one or more quality control metrics associated with error level values or value ranges exceeding the error level definition.   
     
     
         19 . The method of  claim 18 , wherein selecting the one or more quality control metrics associated with the error level values or the value ranges exceeding the error level definition comprises:
 querying a lookup table based on the error level definition.   
     
     
         20 . The method of  claim 17 , wherein selecting the sequence of quality control metrics based on the error level definition representing the first quantization error value comprises:
 selecting one or more quality control metrics based on characteristics of the image or of the transform coefficients.