Optimized prediction based image compression
View Patent ↗A technique for optimizing a prediction method of samples in blocks of an image is provided. The image includes a first block, a second block, a third block, and a fourth block, each of the blocks include 8×8 blocks and form one Macro block. The method includes performing a prediction of the second block, the third block and the fourth block by performing at least one of prediction methods. A prediction error per block (P e ) is computed for each prediction method that is performed to predict the second block, the third block, and the fourth block. The prediction error per block (P e ) equals an original block value minus a predicted block value. An optimal prediction method is chosen from each of the prediction methods performed that results in minimum Σ|P e | pixels per block (summation on pixels per block). The optimal prediction method and the P e for each block are stored.
1. A method of optimizing a prediction method of samples in blocks of an image, said image comprising a first block, a second block, a third block, and a fourth block, wherein said first block, said second block, said third block, and said fourth block comprise 8×8 blocks, wherein each of said 8×8 blocks form one Macro block, said method comprising:
performing a prediction of said second block from said first block by performing any of a DC prediction method and a horizontal prediction method;
performing a prediction of said third block from said first block and said second block by performing said DC prediction method, and at least one of a vertical prediction method and a vertical left prediction method;
performing a prediction of said fourth block from said first block, said second block, and said third block by performing any of said DC prediction method and said horizontal prediction method or any of said vertical prediction method and a vertical right prediction method;
computing a prediction error per block (P e ) using a computer processor for each prediction method that is performed to predict said second block, said third block, and said fourth block, wherein said prediction error per block (P e ) equals an original block value minus a predicted block value;
choosing an optimal prediction method from each of the prediction methods performed, wherein said performed prediction methods comprise said DC prediction method, said horizontal prediction method, and at least one of said vertical prediction method, said vertical left prediction method, and a vertical right prediction method; and
storing said optimal prediction method on a memory device.
2. The method of claim 1 , wherein said choosing said optimal prediction method comprises:
computing a summation of said prediction error per block for each said DC prediction, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method, wherein said summation equals Σ|P e | (summation on pixels per block); and
choosing said optimal prediction method from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method based on said summation that results in a minimum Σ|P e | (summation on pixels per block).
3. The method of claim 1 , wherein said choosing said optimal prediction method comprises:
passing said P e for each said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method through a Discrete Cosine Transformation, a Zigzag scan, and a Huffman Coding; and
choosing said optimal prediction method from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method that results in a minimum number of output bits.
4. The method of claim 1 , further comprising storing said optimal prediction method that results in a minimum number of output bits.
5. The method of claim 1 , further comprising storing said P e for each of said first block, said second block, said third block, and said fourth block, wherein said P e stored comprises a JPEG encoded P e .
6. The method of claim 1 , further comprising performing a prediction of a fifth block based on a first diagonal block, a second diagonal block, a third diagonal block, a first vertical block, and a horizontal block.
7. The method of claim 1 , wherein said prediction method comprises a luminance intra-block prediction method.
8. A method of optimizing a prediction method of samples in blocks of an image, said image comprising a first block, a second block, and a third block, wherein said first block, said second block, and said third block comprise 16×16 blocks, wherein each of said 16×16 blocks form one Luma Macro block, each of said Luma macro block corresponding to an 8×8 color block, said method comprising:
performing a prediction of said second block from said first block by performing any of a DC prediction method and a horizontal prediction method;
performing a prediction of said third block from said first block, said second block, and said third block by performing any of a DC prediction method and any of a vertical prediction method and a vertical left prediction method;
performing a prediction of a fourth block from said first block, said second block, and said third block by performing any of said DC prediction method and said horizontal prediction method or any of said vertical prediction method and a vertical right prediction method;
computing a prediction error per block (P e ) using a computer processor for each prediction method that is performed to predict said second block, said third block, said fourth block, wherein said prediction error per block (P e ) equals an original block value minus a predicted block value;
choosing an optimal prediction method from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method; and
storing said optimal prediction method on a memory device.
9. The method of claim 8 , wherein said choosing said optimal prediction method comprises:
computing a summation of said prediction error for all pixels per block for each said DC prediction method, said horizontal predication method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method, wherein said summation equals Σ|P e |; and
choosing said optimal prediction method from each of the prediction methods performed, wherein said performed prediction methods comprise said DC prediction method, said horizontal prediction method, and at least one of said vertical prediction method, said vertical left prediction method, and a vertical right prediction method.
10. The method of claim 9 , wherein said choosing said optimal prediction method comprises:
passing said P e for each said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method through a Discrete Cosine Transformation, a Zigzag scan, and a Huffman Coding; and
choosing said optimal prediction method from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method that results in a minimum number of output bits.
11. The method of claim 9 , further comprising storing said optimal prediction method that results in a minimum number of output bits.
12. The method of claim 9 , further comprising storing said P e for each of said first block, said second block, said third block, and said fourth block, wherein said P e stored comprises a JPEG encoded P e .
13. The method of claim 12 , wherein said prediction method comprises a color intra-block prediction method.
14. A system optimizing a prediction method of samples in blocks of an image, said image comprising a first block, a second block, a third block, and a fourth block, wherein said prediction method comprises at least one of a DC prediction method, a horizontal prediction method, a vertical prediction method, a vertical left prediction method, and a vertical right prediction method, said system comprising:
means for performing a prediction of said second block;
means for performing a prediction of said third block;
means for performing a prediction of said fourth block;
means for computing a prediction error per block (P e ) for each of said prediction methods that are performed to predict said second block, said third block, said fourth block, wherein said prediction error per block (P e ) equals an original block value minus a predicted block value;
means for choosing an optimal prediction method from each of the prediction methods performed;
means for storing said an optimal prediction method; and
means for storing said prediction error per block (P e ) for said blocks,
wherein said prediction of said second block from said first block is performed by any of a DC prediction method and a horizontal prediction method,
wherein said prediction of said third block from said first block and said second block is performed by any of said DC prediction method and any of said vertical prediction method and said vertical left prediction method, and
wherein prediction of said fourth block from said first block, said second block, and said third block is performed by any of said DC prediction method and said horizontal prediction method or said vertical prediction method and any of a vertical left prediction method and a vertical right prediction method.
15. The system of claim 14 , further comprising means for computing a summation of said prediction error for all pixels per block for each said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method, wherein said summation equals Σ|P e | (summation on pixels per block).
16. The system of claim 15 , wherein an optimal prediction method is chosen from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method based on said summation that result in a minimum Σ|P e | (summation on pixels per block).
17. The system of claim 14 , further comprising means for performing a Discrete Cosine Transformation, a Zigzag scan, and a Huffman Coding on each of said prediction error per block (P e ).
18. The system of claim 17 , wherein said optimal prediction method is chosen from said DC prediction method, said horizontal prediction method, said vertical prediction method, said vertical left prediction method, and said vertical right prediction method based on each of said prediction error per block (P e ) that results in a minimum number of output bits.
19. The system of claim 14 , wherein said P e stored comprises a JPEG encoded P e .
20. The system of claim 14 , wherein said prediction method comprises at least one of a luminance intra-block prediction method, or a color intra-block prediction method.