Image processing apparatus for performing adaptive data processing in accordance with kind of image

• US 5,872,864 A
• Filed: 11/27/1995
• Issued: 02/16/1999
• Est. Priority Date: 09/25/1992
• Status: Expired due to Fees

• Current Assignee: Olympus Optical Corporation Limited (Olympus Corporation)
• Original Assignee: Olympus Optical Corporation Limited (Olympus Corporation)
• Inventors: Tatsuta, Seiji, Imade, Shinichi
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): Prikockis, Larry J.
• Application Number:
  US08/562,637
• Time in Patent Office: 1,177 Days
• Field of Search: 382/173, 382/176, 382/270, 382/271, 382/272
• US Class Current: 382/176
• CPC Class Codes: G06T 9/002 using neural networks
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables
• CPC Class Codes: H04N 1/40062 Discrimination between different image types, e.g. two-tone, continuous tone

Abstract

An image processing apparatus includes an image input section, a color image/monochrome image converting section for performing image area division, a binarization circuit for binarizing a converted monochrome image, a reducing section for reducing a binary image, a boundary extracting section for extracting the boundaries between the areas of constituent elements constituting an input image, e.g., a binary image and a continuous gradation image, and a kind-of-image determining section for determining the kinds of images in partial areas defined by the extracted boundaries. For example, the image processing apparatus further includes a data compressing section. In the image processing apparatus, pre-processing is performed to efficiently performing pattern determination in consideration of localization of frequencies of occurrence of edges and black pixel patterns of an image, and kind-of-image determination is performed by a neural network on the basis of the data obtained by pre-processing, thus performing suitable processing such as data compression in accordance with the determination result.

Claims

1. An image processing apparatus comprising:

image input means for providing image data including various kinds of information;

binarization means for converting the image data provided by said image input means into binary image data;

reducing means for converting the binary image data provided by the binarization means into reduced image data used for area extraction;

constituent element area extracting means for extracting, from the reduced image data provided by the reducing means, an area in which image data having equivalent attributes are continuous;

kind-of-image determining means for determining a kind of an image included in the binary image data provided by said binarization means on the basis of an estimation criterion generated by use of statistical data that represents a feature amount of each of a plurality of kinds of images extracted beforehand, said kind-of-image determining means determining the kind of an image for each of said areas extracted by said constituent element area extracting means;

data compression scheme selecting means for selecting a suitable data compression scheme in accordance with the area extracted by said constituent element area extracting means and the kind of image determined by said kind-of-image determining means; and

data compressing means for compressing image data from said image input means by the compression scheme selected by said data compression scheme selecting means;

wherein said kind-of-image determining means comprises;

a main component vector calculating section for statistically sampling the binary image data from said binarization means a plurality of times for every predetermined number of blocks from an area corresponding to the extracted area of image data output by the image input means, and converting binary image data expressed in units of blocks by normalized orthogonal vectors of pixels constituting a block into uncorrelated data, to produce main component vectors;

an inner product calculating section for calculating, in units of blocks, inner products of the main component vectors using partial image data of the block as a vector sampled from said image every block having a predetermined size; and

a neural network section for performing preliminary learning on the basis of teacher data which is input in units of block data manually sampled, to set the number of the kinds of images to be determined from vectors which represent feature amounts defined as inner products of said main component vectors from said inner product calculating section.

2. An apparatus according to claim 1, wherein said kind-of-image determining means comprises:

another inner product calculating section for calculating inner products of the binary image data from said binarization means and said main component vectors in units of blocks; and

another neural network section for determining a kind of image by using a network constructed by preliminary learning information, from outputs from said other inner product calculating section.

3. An image processing apparatus comprising:

image input means for providing image data having various kinds of information;

binarization means for converting image data provided by said image input means into binary image data;

reducing means for converting the binary image data provided by the binarization means into reduced image data used for area extraction;

constituent element area extracting means for extracting, from the reduced image data provided by the reducing means, an area in which image data having equivalent attributes are continuous;

kind-of-image determining means for determining a kind of an image included in the binary image data provided by said binarization means, on the basis of an estimation criterion generated by use of statistical data that represents a feature amount of each of a plurality of kinds of images extracted beforehand, said kind-of-image determining means determining the kind of an image for each of said areas extracted by said constituent element area extracting means;

data compression scheme selecting means for selecting a suitable data compression scheme in accordance with the area extracted by said constituent element area extracting means and the kind of image determined by said kind-of-image determining means; and

data compressing means for compressing image data from said image input means by the compression scheme selected by said data compression scheme selecting means;

wherein said kind-of-image determining means extracts a feature reflecting a difference in image kind from statistical data and determines an estimation criterion used at the time of image kind determination, said kind-of-image determining means comprising;

a KL main component vector calculating section for statistically sampling the binary image data from said binarization means a plurality of times for every predetermined number of blocks, and converting binary image data expressed in units of one block by normalized orthogonal vectors of pixels constituting a block into uncorrelated data;

a base vector extracting section for calculating base vectors with respect to the binary image data converted into the uncorrelated data by said main component vector calculating section;

an inner product calculating section for calculating, in units of one block, inner products of the base vectors calculated by said base vector extracting section and the binary image data from said binarization means; and

a neural network section for performing preliminary learning on the basis of teacher data input in units of one piece of block data manually sampled to set the number of the kinds of images to be determined from vectors representing feature amounts as inner products from said inner product calculating section.

4. An apparatus according to claim 3, wherein said kind-of-image determining means determines an actual image kind on the basis of the determined estimation criterion, said kind-of-image determining means further comprising:

a second inner product calculating section for calculating inner products of the binary image data from said binarization means and the base vectors obtained in advance in units of one block; and

a second neural network section for determining a kind of image by using a network constructed by preliminary learning with respect to outputs from said inner product calculating section.