基于改进低秩稀疏分解的光斑图像降噪方法研究

Research on Improved Low-rank and Sparse Decomposition-Based Method for Spot Images Denoising

为了提升光斑质心定位精度,提出了一种基于改进低秩稀疏分解(Low-rank and sparse decomposition,LRSD)的图像降噪方法。以增强LRSD降噪能力为目标,在加权核范数最小化(Weighted nuclear norm minimization,WNNM)模型基础上引入全变差(Relative total variation,RTV)范数构建了RTV-WNNM模型,并采用交替方向乘数法(Alternating direction multiplier method,ADMM)对该凸问题进行求解;以增强LRSD细节保持能力为目标,分别引入免疫扰动和退火策略改进了粒子群优化算法(Particle swarm optimization,PSO),并用于LRSD低秩奇异值阈值自适应选取,从而形成一种兼顾噪声去除与局部维系的图像降噪融合算法。仿真和试验结果表明,提出的方法相比于非局部均值滤波(Non-local mean filtering,NLMF)、BM3D等当前主流算法具有更好的降噪效果,显著地提升了含噪光斑图像质心定位精度。

In order to improve the accuracy of centroid location for noise-containing spot images,an image denoising method based on low-rank and sparse decomposition(LRSD)is proposed.Relative total variation(RTV)is introduced into the method on the basis of the weighted nuclear norm minimization(WNNM)model to construct a new RTV-WNNM model,so as to enhance the denoising capability of LRSD method.Alternating direction multiplier method(ADMM)is used to solve this convex problem iteratively.Moreover,to enhance the detail-preserving capability of LRSD denoising method,the immune disturbance and annealing strategy is introduced respectively to improve particle swarm optimization(PSO)algorithm,and is applied to LRSD singular value threshold adaptive selection.Furthermore,it can form an image denoising fusion algorithm,which has both denoising and detail-preserving capabilities.Simulation and experimental results show that the proposed method can reach better results in both quantity measure and visual quality than the state-of-the-art image denoising methods such as non-local mean filtering(NLMF)and BM3D,and it has also remarkably improved the accuracy of centroid location for noise-containing spot images.