基于自适应平方变换的工件去噪方法

Workpieces denoising based on adaptive square transform

针对噪声影响工件图像分割、跟踪等问题,给出一种基于自适应平方变换方法.首先将噪声图像中提取的噪声图像块减去块均值,固定稀疏水平,学习平方变换,更新稀疏水平,作为下一次学习平方变换的稀疏水平,然后更新迭代学习平方变换和稀疏水平,最后一次迭代的去噪块的均值估计作为去噪图像.实验结果表明,给出的方法能较好地滤除噪声.与核奇值分解(K-SVD,kernal singular value decompostion)算法相比,该算法去噪后图像的峰值信噪比(PSNR,peak signal to noise ratio)约是K-SVD算法的2倍,去噪速度是K-SVD的3.9倍.

In wiew of the problem that noise affects workpiece image segmentation and tracking,in this paper,an approach was proposed to address the problem with adaptive square transform.Firstly,the block of noise images was extracted from the noise image and the mean value of the block was subtracted.Then learning square transform with sparse level was fixed.Updating square transform level was done as the sparse level of the next learning square transform.Finally iteratived learning square transform and sparse level update,and averaged at the denoising blocks′locations in the image generate denoised image estimate.The experimental results show that the proposed method has better denoising performance.Compared with the K-SVD algorithm,the peak signal-to-noise ratio of the denoised image is about twice that of the K-SVD algorithm,and the denoising speed is3.9times that of the K-SVD algorithm.