摘要
针对红外图像处理中的去噪问题,提出了一种基于多尺度几何分析的去噪方法;首先给出了一种改进的模糊阈值选取方法,然后对图像进行bandelets变换,在此过程中对系数进行改进的模糊阈值处理,最终实现图像去噪,同时针对去噪问题给出了bandelets变换过程中压缩率阈值的选取方法;仿真结果表明,对于被加性高斯白噪声污染的图像,该方法的去噪性能要好于正交小波硬阈值去噪方法,并且能够获得很好的边缘保持效果。
Aiming at the problem of infrared image denoising, a new method is given based on muhiscale geometric analysis. At first, the method to compute fuzzy threshold is improved. While making bandelets transform to an infrared image, denoising can be realized by applying the improved fuzzy threshold to bandelets coefficients. At the same time, with respect to bandelets transform, a method to select compression rate threshold is given against denoising. Comparing with hard thresholding based on orthogonal wavelet, the simulation results show that denoising effect of this approach is better for an image that acids additive gauss white noise. Moreover, it can preserve the edges of image well.
出处
《计算机测量与控制》
CSCD
2008年第9期1301-1304,共4页
Computer Measurement &Control
基金
航空科学基金(04I53067)
航天科技创新基金资助项目(05C53005)
