基于复无下采样轮廓波和Gaussian小波支持向量回归的红外目标图像背景抑制

Background Suppression of Small Infrared Target Image Based on Nonsubsampled Complex Contourlet Transform and Gaussian Wavelet Support Vector Regression

针对存在背景干扰和噪声情况下的红外目标图像背景抑制问题,提出了一种基于复无下采样轮廓波变换(NSCCT)和Gaussian小波支持向量回归(SVR)的背景抑制方法。该方法对红外目标图像进行NSCCT,然后根据其系数的相关特性去噪,从而抑制了大部分背景杂波;采用Gaussian小波SVR对去噪后的红外目标图像进行处理得到预测图像,并用去噪后图像减去预测图像得到残差图像,即背景抑制结果。针对红外目标图像进行了大量实验,并与近年来提出的3种背景预测方法,即基于最小二乘支持向量回归(LS-SVR)、基于SVR及基于最小二乘的红外目标图像背景抑制方法进行了比较,结果表明所提出的方法去噪效果好,背景抑制性能更优。

For the interference and no background suppression problem of dim target infrared image that contains background ise, a new background suppression method based on nonsubsampled complex contourlet transform （NSCCT） and Gaussian wavelet support vector regression （SVR） is presented. With this method, the nonsubsampled complex contourlet transform is performed for the infrared target image, and then the correlation properties of NSCCT coefficients are used to de-noise the image so that the majority of background clutter is suppressed. Gaussian wavelet support vector regression is used to process the denoised infrared image to obtain the predicted image. The predicted image subtracted from the denoised image gives the residual image and the background is suppressed. A large number of experiments are done on infrared images including small targets, and the comparison is made with the background suppression methods of infrared target image based on least squares support vector regression, support vector regression and least squares. The experimental results show that the suggested method can get better denoising result, and the performance of background suppression is superior.