Maximum likelihood estimation of JPEG quantization values
View Patent ↗This invention is a method and apparatus for processing decompressed images. More particularly, this invention relates to methods and apparatus which process images without requiring information on whether the image was previously compressed or how the image was compressed. Quantization values of each image block of a decompressed image are determined a compression method used to compress a decompressed image to be processed is identified. The method and apparatus of this invention is particularly useful to determine if a decompressed image was previously compressed using a JPEG compression technique by retrieving DCT coefficients and determining quantization values to determine original DCT coefficients of a compressed and decompressed image.
1. A method for processing decompressed image data, comprising:
receiving decompressed image data;
creating an estimated quantization table from the received decompressed image data;
processing the decompressed image data based on the created estimated quantization table to form processed electronic image data.
2. The method of claim 1 , further comprising further processing of the decompressed image data without using the created quantization table.
3. The method of claim 1 , wherein creating the estimated quantization table comprises creating the estimated quantization table based on at least one maximum likelihood estimation.
4. The method of claim 3 , further comprising generating the at least one maximum likelihood estimation based on a probability function.
5. The method of claim 3 , further comprising generating the at least one maximum likelihood estimation based on a Gaussian distribution.
6. The method of claim 1 , wherein the decompressed data comprises image data blocks, and creating the estimated quantization table comprises:
determining, for each block, if that block has at least one of truncated image data values or uniform image data values; and
excluding any block having at least one of truncated image data values or uniform image data values.
7. The method of claim 1 , wherein creating the estimated quantization table further comprises generating transformed image data from the decompressed image data using a discrete cosine transform.
8. The method of claim 7 , wherein creating the estimated quantization table further comprises generating a histogram from the transformed image data.
9. The method of claim 8 , wherein creating the quantization table comprises:
identifying a level of a main lobe of the histogram having a highest peak and two adjacent levels of the histogram adjacent to the identified level; and
creating the quantization table based only on the identified and adjacent levels of the histogram.
10. The method of claim 8 , wherein creating the estimated quantization table further comprises rounding each DCT coefficient of the transformed image data.
11. A system for processing decompressed image data, comprising:
a receiver that receives decompressed image data;
a quantization table estimator that creates an estimated quantization table from the received decompressed image data; and
a processor that processes the decompressed image data based on the created estimated quantization table to form processed electronic image data.
12. The system of claim 11 , wherein the processor further processes the decompressed image data without using the created quantization table.
13. The system of claim 11 , wherein the quantization table estimator creates the estimated quantization table based on at least one maximum likelihood estimation.
14. The system of claim 13 , further comprising a maximum likelihood estimator that generates the at least one maximum likelihood estimation based on a probability function.
15. The system of claim 13 , further comprising a maximum likelihood estimator that generates the at least one maximum likelihood estimation based on a Gaussian distribution.
16. The system of claim 11 , wherein:
the decompressed data comprises image data blocks: and
the quantization table estimator comprises a block analyzer that determines, for each block, if that block has one of truncated image data values or uniform image data values, wherein the quantization table estimator excludes any block having at least one of truncated image data values or uniform image data values.
17. The system of claim 11 , wherein the quantization table estimator further comprises a DCT transformer that generates transformed image data from the decompressed image data using a discrete cosine transform.
18. The system of claim 17 , wherein the quantization table estimator further comprises a histogram generator that generates a histogram from the transformed image data.
19. The system of claim 18 , wherein:
the quantization table estimator further comprises a peak identifier that identifies a level of a main lobe of the histogram having a highest peak and two adjacent levels of the histogram adjacent to the identified level; and
the quantization table estimator creates the quantization table based only on the identified and adjacent levels of the histogram.
20. The system of claim 18 , wherein the quantization table estimator further comprises a coefficient rounding circuit that rounds each DCT coefficient of the transformed image data.
21. The method of claim 1 , wherein creating an estimated quantization table from the received decompressed image comprises creating an estimated quantization table from the received decompressed image without having an original quantization table that was used to compress the received decompressed image.
22. The system of claim 11 , wherein the quantization table estimator creates an estimated quantization table from the received decompressed image without having an original quantization table that was used to compress the received decompressed image.