Global threshold method and apparatus

Claim 3

The method of claim 1 wherein the weight signal closest to the pixel data signal is changed by an amount proportional to the difference between the closest weight signal and each pixel data signal.

Claim 4

The method of claim 1 wherein the changes to each weight signals are reduced in proportion to the number of pixel signals compared to each weight signals.

Claim 5

The method of claim 1 further comprising the step of comparing the pixel data signals to a third weight signal, said third weight signal having a value greater than one of the first and second weight signals and less than the other of the first and second weight signals.

Claim 6

The method of claim 4 further comprising the step of increasing each weight signal when one of the weight signals is not updated after a number of comparisons of the pixel data signal to the weight signals.

Claim 7

The method of claim 5 further comprising the step of inactivating one of the weight signals when only one of the weight signals is updated.

Claim 8

The method of claim 1 further comprising the step of generating a series of binary output pixel data signals with a first set of pixel data signals indicative of foreground of the document and a second set of pixel data signals indicative of background of the document.

Claim 9

The method of claim 7 wherein the first set of pixel data signals indicates black text of the document and the second set of pixel data signals indicates white background of the document.

Claim 10

An image signal processor for separating foreground pixel data signals from background pixel data signals comprising: means for scanning a document to generate a series of pixel data signals, each pixel data signal representative of the intensity of the pixel; means for storing the pixel data signals in a memory; means for comparing each pixel data signal to at least first and second weight signals; means for changing the value of the weight signal closest to each of said pixel data signal; means for counting the number of times each weight signal is changed means for averaging the changed weight signals to determine a threshold signal; and means for comparing the pixel data signals to the threshold signal to determine which pixel data signals are greater than the threshold signal and which pixel data signals are less than said threshold signal.

Claim 11

An image signal processor for separating foreground pixel data signals from background pixel data signals comprising: means for scanning a document to generate a series of pixel data signals, each pixel data signal representative of the intensity of the pixel; means for storing the pixel data signals in a memory; means for comparing each pixel data signal to at least first and second weight signals; means for changing the value of the weight signal closest to each of said pixel data signal; means for counting the number of times each weight signal is changed means for averaging the changed weight signals to determine a threshold signal; and means for comparing the pixel data signals to the threshold signal to determine which pixel data signals are greater than the threshold signal and which pixel data signals are less than said threshold signal.

Claim 12

The image signal processor of claim 9 wherein the closest weight signal is changed by an amount proportional to the difference between the closest weight signal and the pixel data signal.

Claim 13

The image signal processor of claim 9 wherein the changes to each weight signals are reduced in proportion to the number of pixel signals compared to each weight signals.

Claim 14

The image signal processor of claim 9 further comprising means for comparing the pixel data signals to a third weight signal, said third weight signal having a value greater than one of the first and second weight signals and less than the other of the first and second weight signals.

Claim 15

The image signal processor of claim 12 further comprising means for increasing each weight signal when one of the weight signals is not updated after a number of comparisons of the pixel data signal to the weight signals.

Claim 16

The image signal processor of claim 13 further comprising means for inactivating one of the weight signals when only one of the weight signals is updated.

Claim 17

An image signal processor comprising: a scanner for generating a series of pixel data signals, each pixel data signal having a magnitude representative of the intensity of the pixel; a memory for receiving and storing each pixel data signal; first and second weight registers for holding first and second digital weight data signals; a first subtractor having one input for receiving the first weight data signal, another input for receiving the pixel data signals, and an output representative of the difference between the two inputs; a second subtractor having one input for receiving the second weight data signal, another input for receiving the pixel data signals, and an output representative of the difference between the two inputs; a comparator for receiving the outputs of each subtractor and having an output indicating the weight register having a stored weight data signal closest in magnitude to the pixel data signal; and a learn rate counter coupled to the weight registers for updating the magnitude of the weight signal in the weight register indicated by the comparator as closest in magnitude to the pixel data signal by changing the magnitude of the weight data signal stored in said weight data signal register.

Claim 18

An image signal processor comprising: a scanner for generating a series of pixel data signals, each pixel data signal having a magnitude representative of the intensity of the pixel; a memory for receiving and storing each pixel data signal; first and second weight registers for holding first and second digital weight data signals; a first subtractor having one input for receiving the first weight data signal, another input for receiving the pixel data signals, and an output representative of the difference between the two inputs; a second subtractor having one input for receiving the second weight data signal, another input for receiving the pixel data signals, and an output representative of the difference between the two inputs; a comparator for receiving the outputs of each subtractor and having an output indicating the weight register having a stored weight data signal closest in magnitude to the pixel data signal; and a learn rate counter coupled to the weight registers for updating the magnitude of the weight signal in the weight register indicated by the comparator as closest in magnitude to the pixel data signal by changing the magnitude of the weight data signal stored in said weight data signal register.

Claim 19

The image signal processor of claim 15 further comprising a counter for counting the number of times each weight register is updated.

Claim 20

The image signal processor of claim 16 further comprising an averaging circuit connected to the weight registers for averaging the weight data signals stored in the registers to provide a threshold data signal.

Claim 21

The image signal processor of claim 17 further comprising a binary image generator having a first input connected to the threshold data signal and a second input serially connected to each of the pixel data signals stored in the memory and having an output signal comprising a series of binary data signals indicating whether the compared pixel data signals are greater than or less than the threshold data signal.

Claim 22

The image signal processor of claim 15 further comprising a third weight register for holding third digital weight data signals; a third subtractor having one input for receiving the third weight data signal, another input for receiving the pixel data signals, and an output representative of the difference between the two inputs; and wherein the comparator comprises an absolute comparator coupled to the three subtactors, said absolute comparator receives the outputs of each subtractor and has an output indicating the weight register having a stored weight data signal closest in magnitude to the pixel data signal.

Claim 23

The image signal processor of claim 19 further comprising means for increasing each weight signal when one of the weight signals is not updated after a number of comparisons of the pixel data signal to the weight signals.

Claim 24

The image signal processor of claim 20 further comprising means for inactivating one of the weight signals when only one of the weight signals is updated.

Claim 25

A global threshold circuit for processing a series of pixel data signals comprising: an input for receiving pixel data signals, first, second and third weight registers for holding first and second and third digital weight data signals; first, second and third subtractors, coupled respectively to said first, second and third weight registers, each subtractor having one input for receiving the respective first, second and third weight data signal, and each subtractor having another input for receiving the pixel data signals, and each subtractor having an output signal representative of the difference between the two inputs; an absolute comparator for receiving the output difference signals of the subtractors and having a comparator output indicating the weight register having a stored weight data signal closest in magnitude to the pixel data signal; and a learn rate counter coupled to the weight registers for updating the magnitude of the weight data signal in the weight register indicated by the comparator as closest in magnitude to the pixel data signal by changing the magnitude of the weight data signal stored in said weight data signal register.

Claim 26

A global threshold circuit for processing a series of pixel data signals comprising: an input for receiving pixel data signals, first, second and third weight registers for holding first and second and third digital weight data signals; first, second and third subtractors, coupled respectively to said first, second and third weight registers, each subtractor having one input for receiving the respective first, second and third weight data signal, and each subtractor having another input for receiving the pixel data signals, and each subtractor having an output signal representative of the difference between the two inputs; an absolute comparator for receiving the output difference signals of the subtractors and having a comparator output indicating the weight register having a stored weight data signal closest in magnitude to the pixel data signal; and a learn rate counter coupled to the weight registers for updating the magnitude of the weight data signal in the weight register indicated by the comparator as closest in magnitude to the pixel data signal by changing the magnitude of the weight data signal stored in said weight data signal register.

Claim 27

The global threshold circuit of claim 22 wherein each weight register comprises a counter for counting the number of times the weight register is updated.

Claim 28

The global threshold circuit of claim 22 further comprising an averaging circuit for receiving the weight data signals of the weight registers and providing an output thresholding signal representative of the average magnitude of the weight data signals.

Claim 29

The global threshold circuit of claim 22 further comprising: first, second and third shift registers coupled respectively to the outputs of the said first, second and third subtractors; and first, second and third adders coupled respectively to the outputs of the first, second and third shift registers, the adders having outputs coupled respectively to the first, second and third weight registers for changing the value of the data signal stored in the respective weight registers wherein said absolute comparator generates an enable signal for one of said weight registers to enable said one register to receive the output of its respective adder.

Claim 30

The global threshold circuit of claim 25 further comprising a learn rate counter coupled to the shift registers for counting the number of pixel data signals processed by the global threshold circuit and for changing the amount of shifting in proportion to the count of said pixel data signals.