Alex Ostrovsky

US Patent:

61284068, Oct 3, 2000

Filed:

Apr 30, 1997

Appl. No.:

8/846549

Inventors:

Alex Ostrovsky - San Francisco CA

Assignee:

Fujitsu Microelectronics, Inc. - San Jose CA

International Classification:

G06K 900
G06K 936

US Classification:

382166

Abstract:

A method of compressing a block of graphics image data into a compressed block operates on uncompressed bitmap image data to generate a compressed representation of the image. It also operates on compressed bitmap image data to generate the original, uncompressed representation. To compress an image, all 8 pixel by 8 pixel blocks of the image are individually processed in sequence. For each block, all distinct colors present in the block are identified and assigned an ordinal number. A data structure for the block is generated to store the number of distinct colors present in the block, a compression image quality mask value, the actual RGBA quadruplet for each color present in the block, and for each distinct color, sets of group and line binary flags indicating the presence of a particular color at a particular pixel within the block. The data structure for the block is then stored in memory or in a computer file and the next block of the image is processed. This process is repeated for all blocks of the bitmap image.

US Patent:

60918500, Jul 18, 2000

Filed:

Sep 24, 1998

Appl. No.:

9/160504

Inventors:

Alex Ostrovsky - San Francisco CA

Assignee:

Fujitsu Microelectronics, Inc. - San Jose CA

International Classification:

G06K 936
G09G 500

US Classification:

382166

Abstract:

A method of compressing and decompressing graphics images. The compression method operates on uncompressed bitmap image data to generate a compressed representation of the image. To compress an image, all 8 pixel by 8 pixel blocks of the image are individually processed in sequence. For each block, all distinct colors present in the block are identified and assigned an ordinal number. Two pixel colors are deemed distinct if the difference between the colors is greater than an epsilon value. The epsilon value is a function of the total brightness of the block. A data structure for the block is generated to store the number of distinct colors present in the block, the actual RGBA quadruplet for each distinct color, and a color ordinal number of each pixel in the block representing the distinct color of the pixel. To decompress the image, the RGBA quadruplets of all pixels within each block are reconstructed from the data structure for the block.