# US 5,351,314 A · Method and apparatus for image enhancement using intensity dependent spread filtering

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 5,351,314 A
- Document ID: US5351314A
- Status: Granted
- Kind: A
- Application number: 07/771,220
- Filed: 1991-10-04
- Granted: 1994-09-27
- Earliest claimed: 1991-10-04
- Assignees: Canon Information Systems, Inc.
- Inventors: Mehrzad R. Vaezi
- Web page: https://patentgrep.com/patent/US5351314A

## CPC

- G06T 5/20: using local operators
- G06V 30/168: Smoothing or thinning of the pattern; Skeletonisation
- G06V 30/10: Character recognition
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Method and apparatus for image enhancement involving smoothing and thinning input image data, segmenting adjacent characters in the smoothed and thinned data, and identifying a segmented character based on a comparison of the segmented character to a dictionary of characters. Smoothing and thinning may be provided by applying to each pixel of input image data a filter whose spread depends on variations in image intensity of surrounding pixels to obtain filtered image data, deriving inverted image data from first difference data of the filtered image data and multiplying the inverted image data and the filtered image data to obtain smoothed and thinned image data. The filters may be Gaussians whose spreads are inversely proportionate to variations in input image intensity plus a constant and the filters may be applied to the image data using a variably filled mask of filter values.

## Claims

### Claim 1

A method of recognizing a character in an image in the form of pixel data, comprising the steps of:
smoothing and thinning the pixel data;
segmenting adjacent characters in the smoothed and thinned data; and
identifying a segmented character based on a comparison of the segmented character to a dictionary of characters, wherein said smoothing and thinning step comprises the steps of
applying to each pixel a filter whose spread depends on variations in image intensity in surrounding pixels to convert inputted image data into filtered image data;
obtaining differential data of the filtered image data;
deriving inverted image data from the differential data of the filtered image data; and
multiplying the inverted image data and the filtered image data to obtain smoothed and thinned image data.

### Claim 2 (depends on 1)

A method according to claim 1, wherein each filter is approximately a Gaussian whose spread is inversely proportionate to a magnitude of variations in image intensity of surrounding pixels plus a constant.

### Claim 3 (depends on 1)

A method according to claim 1, further comprising the step of selecting a mask for each filter and filling a central portion thereof with filter values and a peripheral portion thereof with zeros.

### Claim 4 (depends on 3)

A method according to claim 3, further comprising the step of determining a maximum mask size based on the maximum spread of the filters.

### Claim 5 (depends on 1)

A method according to claim 1, further comprising the step of calculating first difference data of the filtered image data before storing the filtered image data, wherein in said deriving step the first difference data is used for the differential data.

### Claim 6 (depends on 1)

A method according to claim 1, further comprising the step of thresholding the smoothed and thinned image data obtained in said multiplying step.

### Claim 7

An apparatus for recognizing a character in an image in the form of pixel data, comprising:
input means for inputting image data into the apparatus;
means for smoothing and thinning the input image data;
means for segmenting adjacent characters in the smoothed and thinned data; and
a character recognizer for identifying a segmented character based on a comparison of the segmented character to smooth a dictionary of characters, wherein said means for smoothing and thinning comprises:
means for applying to each pixel a filter whose spread depends on variations in image intensity in surrounding pixels to convert inputted image data into filtered image data;
means for obtaining differential data of the filtered image data;
means for deriving inverted image data from the differential data of the filtered image data; and
a multiplier for multiplying the inverted image data and the filtered image data to obtain smoothed and thinned image data.

### Claim 8 (depends on 7)

Apparatus according to claim 7, wherein each filter is approximately a Gaussian whose spread is inversely proportionate to a magnitude of variations in image intensity of surrounding pixels plus a constant.

### Claim 9 (depends on 7)

Apparatus according to claim 7, further comprising means for selecting a mask for each filter and means for filling a central portion of the mask with filter values and a peripheral portion thereof with zeros.

### Claim 10 (depends on 9)

Apparatus according to claim 9, further comprising means for determining a maximum mask size based on the maximum spread of the filters.

### Claim 11 (depends on 7)

Apparatus according to claim 7, further comprising means for calculating first difference data of the filtered image data, wherein said deriving means uses the first difference data for the differential data.

### Claim 12 (depends on 7)

Apparatus according to claim 7, further comprising means for thresholding the smoothed and thinned image data obtained by said multiplier.

### Claim 13 (depends on 7)

Apparatus according to claim 7, wherein said input means includes means for storing binary image data pixel-by-pixel into a memory.

### Claim 14 (depends on 13)

Apparatus according to claim 13, further including means for deriving pseudo-multi-level image data from the binary image data, wherein the pseudo-multi-level image data comprises the input data.

### Claim 15 (depends on 14)

Apparatus according to claim 14, wherein said input means is a scanner.

### Claim 16 (depends on 14)

Apparatus according to claim 14, wherein said input means is a facsimile machine.

### Claim 17 (depends on 7)

Apparatus according to claim 7, further comprising output means for viewing the smoothed and thinned image data.

### Claim 18 (depends on 7)

Apparatus according to claim 7, further comprising an output device for storing the identity of characters recognized by said character recognizer.

### Claim 19

Apparatus for recognizing a character in a pixel image original, comprising:
input means for inputting the pixel image in the apparatus;
means for deriving for each pixel of the pixel image a filter that depends on variations in intensity of surrounding pixels;
means for applying each said filter to the pixel image to obtain a filtered image;
means for obtaining differential data of the filtered image;
means for deriving an inverted image from the differential data of the filtered image;
means for multiplying the inverted image and the filtered image to obtain a smoothed and thinned image;
a character segmenter for segmenting characters in the smoothed and thinned image;
a character recognizer for comparing each segmented character to a character dictionary to identify characters; and
a memory for storing the identity of the identified characters.

### Claim 20 (depends on 19)

Apparatus according to claim 19, further comprising output means for outputting the smoothed and thinned image.

### Claim 21 (depends on 19)

Apparatus according to claim 19, wherein the pixel image input by said input means comprises a binary pixel image, and further including means for converting the binary pixel image into a pseudo-multi-level pixel image.

### Claim 22 (depends on 21)

Apparatus according to claim 21, further comprising means for information-processing the characters stored in said memory, and an output device for outputting the information-processed characters.

### Claim 23 (depends on 22)

Apparatus according to claim 22, wherein said input means comprises a facsimile machine and wherein said output device is adapted to convert the information-processed characters into a facsimile-compatible format.

### Claim 24 (depends on 19)

Apparatus according to claim 19, further comprising means for permitting selective actuation of said means for deriving a filter for each pixel, said means for applying, said means for deriving an inverted image, and said means for multiplying.

### Claim 25

Apparatus for recognizing a character in a pixel image, comprising:
input means for inputting the pixel image to the apparatus;
a first memory for storing process steps that (a) derive for each pixel of the pixel image a filter that depends on variations in intensity of surrounding pixels, (b) apply each said filter to the pixel image, (c) obtain differential data of the filtered image, (d) derive an inverted image from the differential data of the filtered image, and (e) multiply the inverted image and the filtered image so as to obtain a smoothed and thinned image;
a processor cooperating with said first memory to execute the process steps stored in said first memory so as to form a smoothed and thinned image from the pixel image;
a second memory for storing the smoothed and thinned image formed by said processor; and
means for determining the character represented by the smoothed and thinned image stored in said second memory.

### Claim 26 (depends on 25)

Apparatus according to claim 25, further comprising output means for outputting the smoothed and thinned images.

### Claim 27 (depends on 25)

Apparatus according to claim 25, wherein the pixel image inputted by said input means comprises a binary pixel image, and wherein said first memory further includes process steps to convert the binary pixel image into a pseudo-multi-level pixel image.

### Claim 28 (depends on 27)

Apparatus for claim 27, further comprising (i) a character memory for storing characters, (ii) means for information-processing the characters stored in said character memory, (iii) and an output device for outputting the information-processed characters.

### Claim 29 (depends on 28)

Apparatus according to claim 28, wherein said input means comprises a facsimile machine and wherein said output device is adapted to convert the information-processed characters into a facsimile-compatible format.

### Claim 30 (depends on 25)

Apparatus according to claim 25, further comprising process steps to permit selective actuation of said process steps to derive a filter for each pixel, apply each filter, derive an inverted image, and multiply. PG,39