基于人工神经网络的彩色摄像系统颜色信号失真的校正方法

Method of Calibration about Color Information Distortion of Color Camera System by Neural Networks

本文为了提高彩色摄像系统对颜色信息的校正质量,改善彩色重现的效果,提出了一种基于人工神经网络的色彩校正方法。该方法利用神经网络的非线性变换的特点,实现RGB→XYZ变换或XYZ→RGB的变换。并对网络的学习、变换误差进行了分析。如采用硬件的神经网络,可望实现实时校正。

In this paper, to make raising the quality of color camera system for calibrating color information and improving effect of color reappear, we proposes a new method of color calibration by means of artifical neural networks(ANN), which takes advantage of the non linear conversion in ANN to make the transforming: R G B→X Y Z or X Y Z→R G B. And we analysed to learning of networks and transforming error. The realtime calibration can be made if use hard networks.