| US 7,457,458 B1 | ||
| Method and apparatus for defining and correcting image data | ||
| Florian Daniel, Magdeburg (Germany); Gerald Krell, Magdeburg (Germany); and Bernd Michaelis, Biederitz (Germany) | ||
| Assigned to INB Vision AG., Magdeburg (Germany) | ||
| Filed on Nov. 27, 2000, as Appl. No. 9/723,540. | ||
| Claims priority of application No. 199 56 842 (DE), filed on Nov. 26, 1999; and application No. 100 22 009 (DE), filed on May 05, 2000. | ||
| Int. Cl. G06K 9/00 (2006.01); G06K 9/62 (2006.01) | ||
| U.S. Cl. 382—159 [382/156; 382/162] | 4 Claims |
| 1. An apparatus for defining and at least partially correcting errors of an image reproduction system, said errors being deviations
between an image of predetermined quality and its reproduction, said errors being caused by defects related to color channels
in the image reproduction system, comprising:
image inputting means feeding N2 neuronal nets for N color channels, with output of the neuronal nets being added and connected to inputs of an image reproduction
device for correction of said N color channels;
neuronal nets spatially coupling pixel values of different color channels of image data to be reproduced by mono-layered neuronal
networks via space-variant weights, implemented on a predetermined circuit, comprising parameters established by a learning
process on the basis of a reproduced test image of predetermined quality,
a storage for image data to be reproduced that connects to inputs of the neuronal nets and feeds the pixel values of different
color channels to the inputs of the neuronal nets as target data; and
an image recording device, for generating digital data of an uncorrected image of a test image provided by the image reproduction
device, connected to the inputs of the neuronal nets during the learning process for defining the parameters of the neuronal
nets.
|