Preliminary Research Of Sub-regional Color Correction Algorithm Based On Four-Band Image

IR cameras are used in some optical engineering, but imaging devices do not have the function of color constancy. The visible light and most of the near infrared light can be responded by the CCD or CMOS image sensor. During the day time, the picture deviates from the normal image because of the near infrared light. To eliminate the interference of the infrared light, the main filtered method of infrared filter algorithm is using the selection of optimal conversion opportunity on the basis of the traditional mechanical devices. This method can extend the life of mechanical devices, but the failure rate is high. To substitute the traditional mechanical switching device, this paper proposes a method that we can use bimodal filters to filter out the most IR except for the850nm and apply the infrared filter algorithm to the rest of850nm infrared light. The disadvantage of this method is that the image has varying degrees of color cast due to the different target object or the time. So finding out the right color correction algorithm and restoring original true color, is the premise of this paper.In this paper, we propose a four-band image color correction algorithm to solve the problems that traditional color correction algorithm is not ideal in four-band image, and the850nm IR has different effects on the trees and background.By this way, we can use threshold segmentation method to separate the trees and background. And then, R and B channel values of the two regions is adjusted according to the principles of color balance. Experimental results show that the method can restore original true color in most target areas of the four-band image. Besides that,the algorithm is relatively simple and lays the foundation for subsequent research on video monitoring system.