Selective document image data compression technique

• US 5,754,697 A
• Filed: 05/15/1995
• Issued: 05/19/1998
• Est. Priority Date: 12/02/1994
• Status: Expired due to Fees

• Current Assignee: Lawrence Livermore National Security LLC (Government of the United States of America)
• Original Assignee: Regents of the University of California (University of California)
• Inventors: Petrich, Loren I., Fu, Chi-Yung
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Do, Anh Hong
• Application Number:
  US08/441,152
• Time in Patent Office: 1,100 Days
• Field of Search: 382/232, 382/174, 382/229, 382/289, 382/269, 382/261, 348/364, 348/415, 348/411, 348/407, 348/416, 358/506, 358/428
• US Class Current: 382/232
• CPC Class Codes: G06T 9/004 Predictors, e.g. intraframe, interframe coding
• CPC Class Codes: G06T 9/007 Transform coding, e.g. discrete cosine transform

Abstract

A method of storing information from filled-in form-documents comprises extracting the unique user information in the foreground from the document form information in the background. The contrast of the pixels is enhanced by a gamma correction on an image array, and then the color value of each of pixel is enhanced. The color pixels lying on edges of an image are converted to black and an adjacent pixel is converted to white. The distance between black pixels and other pixels in the array is determined, and a filled-edge array of pixels is created. User information is then converted to a two-color format by creating a first two-color image of the scanned image by converting all pixels darker than a threshold color value to black. All the pixels that are lighter than the threshold color value to white. Then a second two-color image of the filled-edge file is generated by converting all pixels darker than a second threshold value to black and all pixels lighter than the second threshold color value to white. The first two-color image and the second two-color image are then combined and filtered to smooth the edges of the image. The image may be compressed with a unique Huffman coding table for that image. The image file is also decimated to create a decimated-image file which can later be interpolated back to produce a reconstructed image file using a bilinear interpolation kernel.--(235 words)

Claims

1. A method of storing documentary information, the documentary information comprising background information and user-recorded information in a digitized image array comprising a plurality of pixels each having a color value, comprising:

extracting the user information from the background information;

converting the user information to a two-color format;

compressing the user information;

wherein the extracting includes enhancing the contrast of the pixels by performing a gamma correction on the array;

enhancing the color value of each of said pixels;

identifying pixels of said image array being edge pixels lying on edges of said image;

converting pixels at an edge to a black color value and an adjacent pixel to a white color value, respectively;

determining a distance between each black pixel in the array from other black pixels in the array, and converting each of two pixels that exceed a given separation distance between themselves to a white color value;

creating a filled edge array of pixels each corresponding to a respective pixel in said image array.

2. The method of claim 1 wherein said step of converting comprises the sub-steps of:

creating a first two color image of the scanned image by converting all pixels over a threshold color value to black and all pixels below the threshold color value to white;

creating a second two color image of the filled edge file by converting all pixels over a second threshold value to black and all pixels below the second threshold color value to white;

combining the first two color image and the second two color image; and

filtering the combined image to smooth the edges of the image.

3. The method of claim 2 wherein said step of compressing comprises:

compressing said combined image using a compression algorithm with an optimized Huffman table generated for each image compressed and that is based on the composition of said image, wherein a customized Huffman table for each document in a set of images may then be transmitted or stored with said image and used for decompression, and provides for a decompression of data that is relatively computationally non-intensive.

4. A method for processing an image represented by an image array of pixels, each pixel having a color value and a position, comprising the steps of:

enhancing contrast of the pixels by performing a gamma correction on each pixel in the array;

enhancing the color value of each of said pixels by iteratively solving for a new color value based on the position of the pixel in the array;

identifying pixels of said image array being edge pixels lying on edges of said image;

converting sets of adjacent edge pixels to a black color value and a white color value, respectively;

determining a distance between each black pixel in the array and neighboring pixels of any color in the array, and, if there are no nearby pixels within a window of directions, converting said black pixels to white ones;

creating a filled edge array of pixels each corresponding to a respective pixel in said image array;

creating a two-color image of the filled edge array by converting all pixels darker than a first threshold color value to black and all pixels lighter than the threshold color value to white;

creating a two-color image of the scanned image array by converting all pixels darker than a second threshold color value to black and all pixels lighter than the second threshold color value to white;

comparing the two-color image of the scanned image array with the two-color image of the filled edge array to create a combined image;

filtering the combined image to smooth the edges of the image; and

compressing the filtered, combined image.

5. A method, comprising the steps of:

providing an image array of pixels to be compressed;

performing a gamma correction on the color value for each pixel in the array;

enhancing a color value for each pixel in the array;

identifying pixels of said image array being edge pixels lying on edges of said image;

performing a two-color reduction on the edge pixels;

performing a line length correction on all black pixels in the array;

creating a filled edge array of pixels each corresponding to a respective pixel in the image array representing said subject image;

creating a two-color image of the filled edge array by converting all pixels darker than a first threshold color value to black and all pixels lighter than the threshold color value to white;

creating a two-color image of the scanned image array by converting all pixels darker than a second threshold color value to black and all pixels lighter than the threshold color value to white;

comparing the two-color image of the scanned image array with the two-color image of the filled edge array to create a combined image;

filtering the combined image to smooth the edges of the image; and

compressing the filtered, combined image to form a compressed image file.

6. The method of claim 5 further including the steps of:

transmitting the compressed image file from a first system to a second system; and

de-compressing the compressed image file.

7. The method of claim 5 further including the steps of:

storing the compressed image file from a first system to a second system; and

de-compressing the compressed image file.

8. The method of claim 6 further including the step of:

filtering the de-compressed image file to smooth the edge pixels.

9. A method of storing documentary information, the documentary information comprising background information and user-recorded information in a digitized image array comprising a plurality of pixels each having a color value, comprising:

extracting the user information from the background information;

converting the user information to a two-color format;

compressing the user information;

wherein the extracting includes enhancing contrast of the pixels by performing a gamma correction or some similar intensity remapping on each pixel in the array;

enhancing the color value of each of said pixels by iteratively solving for a new color value based on the position of the pixel in the array;

identifying pixels of said image array being edge pixels lying on edges of said image;

converting each edge pixel to a black color value and an adjacent pixel to a white color value;

determining a distance between each black pixel in the array and neighboring pixels of any color in a number of directions in the array, and, if there are no nearby pixels within a set of directions, converting said black pixels to white ones; and

creating a filled edge array of pixels each corresponding to a respective pixel in said image array.

10. The method of claim 9 wherein the step of converting the information to a two color format comprises:

creating a two-color image of the filled edge array by converting all pixels darker than a first threshold color value to black and all pixels lighter than the threshold color value to white;

creating a two-color image of the scanned image array by converting all pixels darker than a second threshold color value to black and all pixels lighter than the second threshold color value to white; and

comparing the two-color image of the scanned image array with the two-color image of the filled edge array to create a combined image.

11. The method of claim 9 wherein the step of compressing the user information comprises:

compressing said combined image using a compression algorithm with an optimized Huffman table generated for each image compressed and that is based on the composition of said image, wherein a customized coding table for each document in a set of images may then be transmitted or stored with said image and used for decompression, and that provides for a decompression of data that is relatively computationally non-intensive.

12. The method of claim 11 wherein the step of compressing the user information comprises:

defining a Huffman coding table for each combined image.

13. An apparatus for compressing an image file, comprising:

means for decimating the image file to create a decimated image file (I), the image file comprisig a plurality of image pixels having a dimension n1*n2, the means decimating the image file in each dimension by factors of p1 and p2, the decimated image file having dimensions m1*m2; and

means for interpolating the decimated image file to produce a reconstructed image file (I_res), the means for interpolating using a bilinear interpolation kernel;

wherein the bilinear interpolation kernel is defined as;

##EQU22##

14. The apparatus of claim 13 wherein n1 and n2 are defined as:

n1=(m1-1)*p1

n2=(m2-1)*p2.

15. The apparatus of claim 14 wherein the restored image is given in terms of the reduced image by:

##EQU23##

16. A method of storing documentary information, the documentary information comprising background information and user-recorded information in a digitized image array comprising a plurality of pixels each having a color value, comprising:

extracting the user information from the background information;

converting the user information to a two-color format to create an image file;

compressing the user information by decimating the image array to create a decimated image file, the image array having a dimension n1*n2 and being decimated in each dimension by factors of p1 and p2, the decimated image file having dimensions m1*m2; and

interpolating the decimated image file to produce a reconstructed image file, the means for interpolating using a bilinear interpolation kernel.

17. The method of claim 16, wherein prior to said interpolating step, the method includes the steps of

compressing the decimated image file using JPEG compression;

and

decompressing the decimated image file using JPEG compression.

18. An apparatus for storing documentary information contained in an image file, the image file comprising a digitized image array containing user information and background information, comprising:

compression means for compressing the image file, including

means for extracting the user information from the background information by

enhancing contrast of the pixels by performing a gamma correction on each pixel in the array,

enhancing the color value of each of said pixels by iteratively solving for a new color value based on the position of the pixel in the array,

identifying pixels of said image array being edge pixels lying on edges of said image;

converting each edge pixels to a black color value and an adjacent pixel to a white color value;

determining a distance between each black pixel in the array and neighboring pixels in the array;

creating a filled edge array of pixels each corresponding to a respective pixel in said image array;

creating a two-color image of the filled edge array by converting all pixels darker than a first threshold color value to black and all pixels lighter than the threshold color value to white;

creating a two-color image of the scanned image array by converting all pixels darker than a second threshold color value to black and all pixels lighter than the second threshold color value to white; and

comparing the two-color image of the scanned image array with the two-color image of the filled edge array to create a combined image; and

means for compressing the user information and decompressing the user information, comprising

means for decimating the image file to create a decimated image file (J), the image file comprising a plurality of image pixels having a dimension n1*n2, the means decimating the image file in each dimension by factors of p1 and p2, the decimated image file, the decimated image file having dimensions m1*m2; and

means for interpolating the decimated image file to produce a reconstructed image file (I_res), the means for interpolating using a bilinear interpolation kernel.

19. An apparatus for processing an image-pixel array, comprising:

a first processor, the first processor

enhancing contrast of the pixels by performing a gamma correction on each pixel in the array; and

enhancing the color value of each of said pixels by iteratively solving for a new color value based on the position of the pixel in the array;

a data bus coupled to the first processor;

a second processor, coupled to the data bus and the first processor, the second processor

identifying pixels of said image array being edge pixels lying on edges of said image;

converting each edge pixels to a black color value and an adjacent pixel to a white color value;

determining a distance between each black pixel in the array and neighboring pixels in the array;

creating a filled-edge array of pixels each corresponding to a respective pixel in said image array;

creating a two-color image of the filled-edge array by converting all pixels darker than a first threshold color value to black and all pixels lighter than the threshold color value to white;

creating a two-color image of the scanned image array by converting all pixels darker than a second threshold color value to black and all pixels lighter than the second threshold color value to white; and

comparing the two-color image of the scanned image array with the two-color image of the filled-edge array to create a combined image file.

20. An method for compressing an image file, comprising:

decimating the image file to create a decimated image file (J), the image file comprising a plurality of image pixels having a dimension n1*n2 such that the image file is decimated in each dimension by factors of p1 and p2, the decimated image file having dimensions m1*m2; and

interpolating the decimated image file to produce a reconstructed image file (I_res) using a bilinear interpolation kernel defined as;

##EQU24##

21. The method of claim 20 wherein n1 and n2 are defined as:

n1=(m1-1)*p1

n2=(m2-1)*p2.

22. The method of claim 21 wherein the restored image is given in terms of the reduced image by:

##EQU25##