Network and image area attribute discriminating device and method for use with said neural network

• US 5,884,296 A
• Filed: 03/11/1996
• Issued: 03/16/1999
• Est. Priority Date: 03/13/1995
• Status: Expired due to Term

• Current Assignee: Minolta Co., Ltd. (Konica Minolta Inc.)
• Original Assignee: Minolta Co., Ltd. (Konica Minolta Inc.)
• Inventors: Okamoto, Shinji, Niioka, Makoto, Yamamoto, Shinji, Nakamura, Kazuaki, Itoh, Tetsuya
• Primary Examiner(s): Davis, George B.
• Application Number:
  US08/613,319
• Time in Patent Office: 1,100 Days
• Field of Search: 395/22, 395/23, 382/156, 382/159, 382/160, 382/170, 382/181, 706/20
• US Class Current: 706/20
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/20056 Discrete and fast Fourier transform, [DFT, FFT]
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 7/42 using transform domain methods
• CPC Class Codes: G07D 7/1205 Testing spectral properties
• CPC Class Codes: H04N 1/40062 Discrimination between different image types, e.g. two-tone, continuous tone

Abstract

A device for discriminating an attribute of an image in a block area contained in a document image includes a device for performing a Fourier transformation based on image data in the block area and for determining a spatial frequency spectrum relating to the image in the block area; and a neural network for outputting a discrimination result as to whether or not the attribute of the image in the block area is a halftone dot image based on the spatial frequency spectrum output from the Fourier transform device.

Claims

1. A device for discriminating whether an image in a block area contained in a document image is a halftone dot image, the device comprising:

means for performing a Fourier transformation based on image data in the block area and determining a spatial frequency spectrum relating to the image in the block area; and

a neural network which produces a discrimination result indicating whether or not the image in the block area is a halftone dot image based on the determined spatial frequency spectrum.

2. A device for classifying an image in a block area contained in a document image, the device comprising:

means for calculating a quantity of a feature based on image data in the block area; and

a neural network which produces a discrimination result indicating whether or not the image in the block area is a text image, or a photographic image, or a halftone dot image based on the calculated quantity of the feature.

3. The device as claimed in claim 2, wherein the means for calculating includes:

means for performing Fourier transformation based on image data in the block area and determining a spatial frequency spectrum therefrom; and

means for creating a density matrix based on image data in the block area and determining statistics which describe the density distribution in the block area.

4. A device for classifying an image in a block area contained in a document image, the device comprising:

means for creating a density matrix based on image data in the block area and determining statistics which describe density distribution; and

a neural network which produces a discrimination result indicating whether or not the image in the block area is a text image or a photographic image based on the determined statistics.

5. A neural network for discriminating an attribute of an image based on input image data, comprising:

a first neural network for calculating a quantity of a feature of the image, said first neural network including an input level, an intermediate level and an output level which are trained so that said quantity is calculated at said intermediate level and output values produced at said output level are the same as image values input to said input level; and

a second neural network connected to said intermediate level of said first neural network for outputting an attribute of the image based on the calculated quantity of the feature.

6. The device as claimed in claim 5, further comprising:

means for dividing an image into a plurality of blocks; and

means for inputting image data of the divided blocks, so that an attribute of an image in each block is discriminated.

7. The device as claimed in claim 5, wherein the discriminated attribute is one of a text image, a photographic image, and a halftone dot image.

8. The device as claimed in claim 6, wherein the discriminated attribute is one of a text image, a photographic image, and a halftone dot image.

9. The neural network of claim 5, wherein the image values input to said input level comprise pixel values which respectively describe individual pixels in said image.

10. An attribute discriminating device, comprising:

means for outputting image data of a plurality of blocks having mutually different sizes relative to a target area for which an attribute is to be discriminated within a document image; and

a neural network for discriminating an attribute of the target area based on the image data corresponding to the plurality of blocks.

11. The attribute discriminating device as claimed in claim 10, wherein the neural network includes:

a first neural network for discriminating an attribute of an image in each block based on the corresponding image data from the means for outputting; and

a second neural network for discriminating an attribute of an image in the target area synthetically based on results discriminated by the first neural network.

12. The attribute discriminating device as claimed in claim 11, wherein the means for outputting includes means for compressing image data in the plurality of blocks so as to reduce an amount of image data.

13. The attribute discriminating device as claimed in claim 12, wherein the means for compressing compresses image data in each block so that an amount of the compressed image data is a certain amount.

14. An attribute discriminating device, comprising:

means for outputting image data in a plurality of blocks having mutually different sizes relative to a target area for which attributes are to be discriminated within a document image;

first means for discriminating an attribute of an image in each block based on the corresponding image data from the means for outputting; and

second means for discriminating an attribute of an image synthetically based on results discriminated by the first means for discriminating.

15. The attribute discriminating device as claimed in claim 14, wherein the second means discriminates an attribute of an image in the target area.

16. The attribute discriminating device as claimed in claim 14, wherein the first means and the second means are constructed by a neural network.