Hybrid neural networks for color identification

Details Hybrid neural networks for color identification Hybrid neural networks for color identification

2000-08-09

2003-05-27

Davis, George B. (Department: 2122)

Data processing: artificial intelligence

Neural network

Learning task

C706S020000, C706S021000, C706S025000

Reexamination Certificate

active

06571228

ABSTRACT:

FIELD OF THE INVENTION
This invention presents a method using a hybrid neural networks including a self-organizing mapping neural network (SOM NN) and a back-propagation neural network (BP NN) for color identification. In the method RGBs ( red, green, blue) of color samples are input as features of training samples and are classified by SOM NN. Afterwards, the outputs of SOM NN are respectively delivered to various BP NN for further learning; and RGBs of the color samples onto corresponding XYZs of IT8. By the way of the above learning structure, a non-linear model of color identification can be set up. After color samples are self-organized-and classified by SOM NN, data can be categorized in clusters as a result of characteristic-difference thereof. Then the data are transmitted to BP NN. This learning system not only can be quickly converged but also lower error discrepancy in operation effectively.
This invention could be widely applied in color quality control and color proof task. No matter in the printing industry, scanner and CCD (charge coupled device) that are image processing devices and must be connected to the above equipment responsible for color quality control and proof task whereby color discrepancy and mal-quality problems can be overcome. In particular, color is not linear data in nature and the discrepancy in brightness and color hue can not be simply solved by linear approaches, or image infidelity can be more serious in practical operation.
SUMMARY OF THE INVENTION
Therefore, the primary object of the present invention is to provide a device a method using a hybrid neural networks, including a self-organizing mapping neural network (SOM NN) and a back-propagation neural network ( BP NN), for RGBs to XYZs color transformation. This technique could apply for color identification, color quality control and color proof task. SOM NN could classify input RGBs to different categories. Each categories of SOM NN will link a BP NN to perform a supervising learning. The learning cycle will stop until reaching a desired inferred XYZs.
Anther object of this invention is to provide a method for color identification using a intelligent system to process color calibration in error tolerance, self-organization and efficiency. This technique reveals a new intelligent solution to solve a color complex problem. This technique actually superiors to a traditional color transformation like a 3×3 matrix.
One further object of the present invention is to provide method for color identification wherein self-learning process can continue in BP NN until a preset termination condition is met in each BP NN for satisfactory color identification.


REFERENCES:
patent: 6314413 (2001-11-01), Otte
patent: 6453246 (2002-09-01), Agrafiotis et al.
patent: 6505181 (2003-01-01), Lambert et al.
Fung et al, “Modular Artificial Neural Network for Prediction of Petrophysical properties from well Log Data”, IEEE Instrumentation and Measurement Technology Conference, Jun. 1996.

Affiliated with

Also associated with

Rating

Hybrid neural networks for color identification does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Hybrid neural networks for color identification, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Hybrid neural networks for color identification will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3041238