论文部分内容阅读
针对存在背景干扰和噪声情况下的红外弱小目标检测问题,提出一种基于无下采样contourlet变换(NSCT)和独立分量分析(ICA)的检测方法。首先原始图像减去通过快速ICA分离出的背景图像,再经NSCT去噪,接着利用新型Top-hat变换滤波得到预处理图像;然后采用基于类内方差及背景与目标面积差的阈值选取方法来分割预处理图像。针对红外小目标图像进行了大量实验,并和基于快速ICA、基于NSCT的红外目标检测方法进行了比较,结果表明所提出的方法抗噪性强,具有更为优越的检测性能。
Aiming at the problem of detecting infrared weak targets in the presence of background noise and noise, a detection method based on nonsubsampled contourlet transform (NSCT) and independent component analysis (ICA) is proposed. Firstly, the original image is subtracted from the background image separated by fast ICA, and then denoised by NSCT. Then the new Top-hat transform filter is used to get the preprocessed image. Then the threshold selection method based on the variance within the class and the difference between background and target area Split the pre-processed image. A large number of experiments were carried out on the infrared small target image and compared with the fast ICA and NSCT based infrared target detection method. The results show that the proposed method is robust against noise and has superior detection performance.