基于奇异值分解和形态滤波的图像噪声抑制算法研究

Study on Image Noise-suppressing Algorithm Based on Singular Value Decomposition and Morphological Filter

通过对在空域上图像去噪问题进行研究,发现一些经典的算法在保留信号细节部分出现了模糊细节或块状现象。针对此问题提出了一种基于奇异值分解和形态滤波的图像噪声抑制算法。首先利用奇异值分解后截取最优特征值得到预处理后的图形,然后利用基于不同大小的可变结构元素的改进的形态滤波函数从不同角度得到各方位去噪图像,结合自适应最优权系数法加权合成处理得到最后的无噪声图像。算法在抑制噪声的同时更好保护了边缘特征信息。理论分析和实验结果表明算法的有效性。

An image denoising method based on singular value decomposition and morphological filter is proposed, owing to the fact that some existing image denoising methods blur details or make the image mosaic. Firstly, the preprocessed image was gotten using optimal threshold method after singular value decomposition, then the improved morphological filter based on different size and direction structuring element was applied to noise-suppressing image of each direction, and the non-noise image was gotten combining the adaptive weighting method. The pro- posed method not only reduced noise but also preserve edge information. Theoretic analysis and experiment result verify the effectiveness of the method.