CGSI pipeline performance improvement

Claim 1

A method for preprocessing an object image prior to input into an image warp transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same row of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a vertical compressed object image size by approximately determining the maximum number of pixels in a single column of said compressed object image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said compressed vertical object image size to said vertical output image size; generating a decimated image that includes only pixels in every Nth row of the compressed object image, wherein the pixels for each Nth row are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; and inputting the input image to the image warp transformation processor.

Claim 2

A method for preprocessing an object image prior to input into an image warp transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same row of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a vertical compressed object image size by approximately determining the maximum number of pixels in a single column of said compressed object image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said compressed vertical object image size to said vertical output image size; generating a decimated image that includes only pixels in every Nth row of the compressed object image, wherein the pixels for each Nth row are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; and inputting the input image to the image warp transformation processor.

Claim 4

A method for preprocessing an object image prior to input into an image warp transformation processor, said warp transformation processor having a vertical warp transformation processor and a horizontal transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same row of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a vertical compressed object image size by approximately determining the maximum number of pixels in a single column of said compressed object image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said compressed vertical object image size to said vertical output image size; generating a decimated image that includes only pixels in every Nth row of the compressed object image, wherein the pixels for each Nth row are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; inputting the input image to the vertical warp transformation processor to produce an intermediate image; computing a horizontal intermediate image size by approximately determining the maximum number of pixels in a single row of said intermediate image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said horizontal intermediate image size to said horizontal output image size; and generating a decimated intermediate image that includes only pixels in every Mth row of said intermediate image; and inputting the decimated intermediate image to the horizontal warp transformation processor.

Claim 5

A method for preprocessing an object image prior to input into an image warp transformation processor, said warp transformation processor having a vertical warp transformation processor and a horizontal transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same row of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a vertical compressed object image size by approximately determining the maximum number of pixels in a single column of said compressed object image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said compressed vertical object image size to said vertical output image size; generating a decimated image that includes only pixels in every Nth row of the compressed object image, wherein the pixels for each Nth row are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; inputting the input image to the vertical warp transformation processor to produce an intermediate image; computing a horizontal intermediate image size by approximately determining the maximum number of pixels in a single row of said intermediate image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said horizontal intermediate image size to said horizontal output image size; and generating a decimated intermediate image that includes only pixels in every Mth row of said intermediate image; and inputting the decimated intermediate image to the horizontal warp transformation processor.

Claim 7

A method for preprocessing object image data prior to input into an image warp transformation processor, said object image data having a plurality of pixels associated therewith which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same column of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a horizontal compressed object image size by approximately determining the maximum number of pixels in a single row of said compressed object image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said compressed horizontal object image size to said horizontal output image size; generating a decimated image that includes only pixels in every Mth column of the compressed object image, wherein the pixels for such Mth column are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; and inputting the input image to the image warp transformation processor.

Claim 8

A method for preprocessing object image data prior to input into an image warp transformation processor, said object image data having a plurality of pixels associated therewith which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same column of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a horizontal compressed object image size by approximately determining the maximum number of pixels in a single row of said compressed object image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said compressed horizontal object image size to said horizontal output image size; generating a decimated image that includes only pixels in every Mth column of the compressed object image, wherein the pixels for such Mth column are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; and inputting the input image to the image warp transformation processor.

Claim 10

A method for preprocessing an object image prior to input into an image warp transformation processor, said warp transformation processor having a vertical warp transformation processor and a horizontal transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same column of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a horizontal compressed object image size by approximately determining the maximum number of pixels in a single row of said compressed object image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said compressed horizontal object image size to said horizontal output image size; generating a decimated image that includes only pixels in every Mth column of the compressed object image, wherein the pixels for such Mth column are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; inputting the input image to the horizontal image warp transformation processor to produce an intermediate image; computing a vertical intermediate image size by approximately determining the maximum number of pixels in a single column of said intermediate image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said vertical intermediate image size to said vertical output image size; and generating a decimated intermediate image that includes only pixels in every Nth column of said intermediate image; and inputting the decimated intermediate image to the vertical warp transformation processor.

Claim 11

A method for preprocessing an object image prior to input into an image warp transformation processor, said warp transformation processor having a vertical warp transformation processor and a horizontal transformation processor, said object image comprised of a plurality of pixels which are arranged in an array of rows and columns, said warp transformation processor transforming said object image into an output image having a known arrangement of rows and columns of pixels different from that of said object image said object image, said method comprising the steps of: grouping consecutive strings of 64 pixels in the same column of the object image into 8 pixel by 8 pixel intermediate pixel blocks, wherein each 8 pixel by 8 pixel block in actuality contains a line segment of the object image; compressing the 8 pixel by 8 pixel intermediate pixel blocks to form a compressed object image; computing a horizontal compressed object image size by approximately determining the maximum number of pixels in a single row of said compressed object image; computing a horizontal output image size by approximately determining the maximum number of pixels in a single row of said output image; calculating a factor M as a function of the ratio of said compressed horizontal object image size to said horizontal output image size; generating a decimated image that includes only pixels in every Mth column of the compressed object image, wherein the pixels for such Mth column are grouped in 8 pixel by 8 pixel blocks of the compressed object image; decompressing the decimated image to produce an input image; inputting the input image to the horizontal image warp transformation processor to produce an intermediate image; computing a vertical intermediate image size by approximately determining the maximum number of pixels in a single column of said intermediate image; computing a vertical output image size by approximately determining the maximum number of pixels in a single column of said output image; calculating a factor N as a function of the ratio of said vertical intermediate image size to said vertical output image size; and generating a decimated intermediate image that includes only pixels in every Nth column of said intermediate image; and inputting the decimated intermediate image to the vertical warp transformation processor.