基于GCV小波阈值去噪的红外弱小目标检测

Dim Small Target Detection Based on Generalized Cross Validation Wavelet Threshold Denoising

针对红外图像弱小目标检测,提出了一种基于GCV小波阈值去噪的检测方法。先将图像进行小波多尺度分解以抑制杂波;然后利用GCV准则,对各个小波系数的子带图像分别进行阈值分割来抑制噪声;最后对降噪后的小波系数进行离散小波反变换,对重构后的图像进行二值分割从而得到最终检测结果。实验结果表明,与传统方法相比,该算法能够很好地抑制背景和噪声,并且对于重建图像能够更加准确地进行目标提取和分割。

Concerning dim small target detection from infrared image ,a new method based on gener -alized cross validation（ GCV） wavelet threshold denoising is presented .Firstly, for an IR image , a wave-let multiscale decomposition method is used to suppress clutter .Then the GCV criteria is applied to calcu-late the GCV threshold for each subband and the images are segmented separately for denoising .Finally, inverse descret wavelet transform （ IDWT） method is used for denoised wavelet coefficients and the target is segmented from the reconstructed image .Experimental results show that the presented algorithm can suppress background and noise well , which detects and segments dim small targets efficiently from recon-structed image .