基于K-SVD字典稀疏分解的实木地板去噪方法

Solid wood flooring denoising based K-SVD dictionary sparse decomposition method

针对实木地板的图像获取过程中,所产生的噪声问题,引入了K-SVD字典的学习算法,提出了一种图像的有用信息稀疏分解去噪的方法,目的是有效的保留实木地板的有用纹理信息,并抑制其中掺杂的噪声。通过对图像稀疏分解后得到的值,来进行图像重构,就可以达到图像的去噪目的。首先,构造一个初始化的DCT字典,对图像分块处理;接着,在这个初始化字典的基础之上,进行纹理信息的稀疏分解,同时,对它们之间的残差值进行奇异值分解,更新字典;最后,利用得出的最优化字典,采用正交匹配重构算法,完成去噪图像的重建。实验表明,该算法得出的图像主观效果好,减少了去噪后的模糊程度及保留更多细节信息,在不同程度的噪声下,PSNR较高。

For the image acquisition process of real wood floor, the noise problem, the introduction of K-SVD dictionary learning algorithm, this paper proposes a useful information of the image sparse decomposition denoising methods, the purpose is to effectively retain the useful texture information of real wood floor, and suppress the noise of the doping.Through to get the value of the after image sparse decomposition, to reconstruct the image, can achieve image denoising.First of all, construct an initialization DCT dictionary, the image block processing; Then, on the basis of the initialized dictionary, sparse decomposition on texture information, at the same time, to among them the residual value of singular value decomposition, update the dictionary; Finally, using it is concluded that the optimization of the dictionary, orthogonal matching reconstruction algorithm, to complete the reconstruction of the image denoising.Experiments show that the algorithm of image subjective effect is good, the fuzzy degree after denoising was decreased, and keep more details information, under the different levels of noise, higher PSNR.