Image analyzer with variable line storage

Claim 3

The system of claim 1 including: neighborhood extraction means having a plurality of registers for temporarily storing the neighborhood windows, said registers being configured in an M and N array having M rows and N columns, with the registers in each row being serially coupled; means for shifting said pixel values through a first row of neighborhood registers and simultaneously loading said RAM with the same pixel values; and means for loading subsequent rows of said neighborhood registers with pixel values from said RAM wherein said RAM serves as a variable line storage device for readily accommodating different raster scan line lengths.

Claim 4

The system of claim 2 wherein said neighborhood extraction registers are configured in a 3×3 array, said RAM temporarily storing about two scan lines of pixel values.

Claim 5

In an image analyzer system having at least one serial neighborhood processing stage for analyzing an image matrix represented by a plurality of raster scan lines of pixel values, said stage having an M row X N Column array of storage registers for temporarily storing a neighborhood window for analysis, said neighborhood window comprising a center pixel and surrounding pixels in the image matrix, wherein the improvement comprises a method of sequentially loading at least first, second and third rows of the neighborhood window registers with successively generated first, second and third raster scan lines of pixel values to thereby successively access substantially all of the neighborhoods in the image matrix, said method comprising: shifting said pixel values through the first row of neighborhood window registers; simultaneously loading a random access memory (RAM) with pixel values of at least the first and second scan lines of the image; thereafter, addressing a pair of memory locations containing at least the first pixel value of the first and second scan lines, and loading said pixel values into the first window registers of the second and third register rows, respectively, as the first pixel value of the third line is entering the first window register of the first register row; analyzing the pixel values contained in the neighborhood window registers; and subsequently shifting the pixel values to adjacent window registers in each row, addressing a pair of memory locations containing successive pixel values of adjacent scan lines, and loading said addressed pixel values into the first window registers of said second and third register rows as new pixel values are shifted into the first window register of the first row to thereby sequentially access substantially all of the neighborhoods in the image for analysis.

Claim 6

In an image analyzer system having at least one serial neighborhood processing stage for analyzing an image matrix represented by a plurality of raster scan lines of pixel values, said stage having an M row X N Column array of storage registers for temporarily storing a neighborhood window for analysis, said neighborhood window comprising a center pixel and surrounding pixels in the image matrix, wherein the improvement comprises a method of sequentially loading at least first, second and third rows of the neighborhood window registers with successively generated first, second and third raster scan lines of pixel values to thereby successively access substantially all of the neighborhoods in the image matrix, said method comprising: shifting said pixel values through the first row of neighborhood window registers; simultaneously loading a random access memory (RAM) with pixel values of at least the first and second scan lines of the image; thereafter, addressing a pair of memory locations containing at least the first pixel value of the first and second scan lines, and loading said pixel values into the first window registers of the second and third register rows, respectively, as the first pixel value of the third line is entering the first window register of the first register row; analyzing the pixel values contained in the neighborhood window registers; and subsequently shifting the pixel values to adjacent window registers in each row, addressing a pair of memory locations containing successive pixel values of adjacent scan lines, and loading said addressed pixel values into the first window registers of said second and third register rows as new pixel values are shifted into the first window register of the first row to thereby sequentially access substantially all of the neighborhoods in the image for analysis.

Claim 7

The method of claim 4 wherein said pixel values are loaded in said memory as a block in sequential address locations.

Claim 8

The method of claim 5 wherein said pixel values stored in the memory are addressed by a pair of recirculating address pointers which are spaced about one scan line length apart.

Claim 9

The method of claim 6 wherein the incoming new pixel values are stored in the same memory locations previously addressed by at least one of said pointers to replace those pixel values loaded into said second and third register rows.

Claim 10

The method of claim 4 wherein said window registers are configured in a 3×3 array.

Claim 11

The method of claim 8 wherein about two successive scan lines are stored in the memory.

Claim 12

The method of claim 4 wherein the stage provides a transformation output for each neighborhood of pixel values contained in the window registers and supplies the output to an input of a substantially identical stage in a pipeline.