基于独立分量分析的近红外图像去噪方法的研究与应用

独立分量分析（Independent Component Analysis,ICA）是一种新的盲源信号分离方法,近年来已吸引了信号处理学界和神经网络学界众多学者的研究兴趣.该文提出了基于独立分量分析的近红外图像去噪方法,该方法能有效地处理用黑白CCD传感器监测电力设备所产生的图像,使这种监测方案具有实用价值.文中简单描述了以黑白CCD传感器为核心器件的电力设备在线监测系统,阐述了ICA的基本原理,介绍了基于负熵判据的FastICA算法,给出了所提方法的具体实现步骤及相应的实验处理结果.结果表明利用该方法可以有效的去除监测图像中的噪声,使温度场的显示更清晰,满足了实际的需求.

RESEARCH AND ITS APPLICATION OF A METHOD OF NEAR-INFRARED IMAGE DE-NOISING BASED ON ICA

Independent component analysis (ICA) is a new approach of blind source separation. In recent years, it has received considerable attention from the signal processing community and the neural network community. A method of near-infrared image de-noising based on ICA is presented in this paper. It can process availably images in monitoring power equipments by using monochrome CCD sensor, and it has practical significance. The on-line monitoring system with a monochrome CCD sensor as the key device is described. The principle of ICA and a FastICA algorithm based on negentropy criterion are introduced. The detailed processes of the method and results are given. The results show it can remove effectively noise of images, which makes the display of temperature field more vivid and it can satisfy the actual requirement.