Feedback control for hybrid compression
AssigneeTexas Instruments Incorporated
InventorSrinivas P. Rao
Life of this patent
Estimating the term…
Abstract
A closed loop feedback system adaptively controls the compression ratio in a Raster Image Processor. The image content is analyzed in real time, and rasterized bitmap is compressed to a sufficient degree to fit into the available frame buffer. This compression may be done by a variety of algorithms depending on image content. The compression ratio is adjusted on the fly by changing the method of compression, more aggressive or selective quantizing of the image, or by a decimation of parts of the image. The algorithms show will execute very efficiently on a Texas Instruments TMS320C82 multiprocessing DSP. Several methods of implementation on the TMS320C82 are shown.
Claims13
A method for adaptively controlling the compression of a bitmap image to guarantee the attainment of a predetermined compression ratio comprising the steps of:
- partitioning said bitmap image into blocks;
- classifying each block according to image content;
- compressing each block by an algorithm selected to be most efficient for the class of the block;
- during compression, comparing the actual attained compression ratio to the target compression ratio and generating an error signal if the target compression ratio has not been attained, where this comparing may be done on a block, row of blocks or page basis and the resultant error signal is used to adjust the compression ratio for subsequent blocks;
- and
- adjusting the compression ratio by changing a block classification threshold in response to said error signal, where compression is increased by shifting the block classification threshold to increase the number of blocks classified as SOLID, and is decreased by increasing the number of blocks classified as IMAGE.
A method of adaptively controlling the compression of a bitmap image to guarantee the attainment of a predetermined target compression ratio comprising the steps of:
- partitioning the bitmap image into blocks;
- classifying each block according to image content type, said step of classifying each block according to image content type including
- determining the alpha of each block as the average absolute variation from the mean of the pixels within the block,
- selecting an alpha threshold,
- classifying a block as SOLID if the alpha of the block is less than the alpha threshold, and
- classifying a block as IMAGE if the alpha of the block is greater than the alpha threshold;
- compressing each block by an algorithm selected to be most efficient for the image content type of the block;
- comparing an actual attained compression ratio to the target compression ratio;
- and
- adjusting the compression ratio for subsequent blocks corresponding to the relationship of the actual attained compression ratio and the target compression ratio, said step of adjusting the compression ratio for subsequent blocks including
- increasing the alpha threshold thereby shifting the block classification to increase the number of blocks classified as SOLID if the actual attained compression ratio is less than the target compression ratio, and
- decreasing the alpha threshold thereby shifting the block classification to increase the number of blocks classified as IMAGE if the actual attained compression ratio is greater than the target compression ratio.
Dependencies are inferred from the claim text.