高分辨率红外图像的分层自适应阈值Curvelet系数萎缩去噪方法

高分辨率红外图像在基于小波系数阈值萎缩的去噪过程中，容易导致边缘模糊或丢失等失真。文中首次引入基于wrapping的第二代快速Curvelet变换，对图像边缘信息进行有效的稀疏保存，并采用分层自适应阈值算法独立估计每个尺度、方向上的Curvelet系数噪声阈值，并针对红外图像的Curvelet系数能量高度集中于低尺度系数的特点，采用尺度相关的硬阈值对染噪图像的Curvelet系数进行处理。实验结果表明：在不同噪声条件下，与基于小波系数的Visu Shrink,Penalized,sparsity-norm阈值等去噪算法相比，文中提出的去噪算法取得了较好的去噪效果，在噪声方差σ=30时，使用该方法的峰值信噪比(PSNR)可高达31.77 dB，去噪后的图像边缘保持良好，具有较好的视觉效果；同时，文中建议算法的计算量比传统Curvelet降低了70%以上，适合在DSP等嵌入式系统应用。

High resolution infrared image denoising based on curvelet and hierarchical adaptive threshold shrinkage

Edge of high resolution infrared image is blurry after denoising with the denoing arithmetic based on wavelet.A new denoising method based on 2th generation curvelet and hierarchical adaptive threshold is proposed to preserve the edge better.The denoising threshold of curvelet coefficient is estimated separately by hierarchical adaptive threshold and the noising image is denoising with hard threshold related to the transform scale because most of energy of curvelet coefficients is concentrated in the low scale coefficients.The experiment results show that the proposed algorithm can obtain a better denoising performance than some arithmetic based on wavelet like VisuShrink,Penalized,sparsity-norm in all kinds of noise spectral density.The PSNR is up to 31.51 dB with the proposed algorithm when noise variance is 30.The images after denoisinig based on curvelet preserve their edge better and is better in visual effect;furthermore,the computational complexity of proposed algorithm decreases 70% compared to classic curvelet transform.