基于L1-L2范数的正则项去噪模型的PCB图像去噪算法研究

Research on PCB image denoising algorithm based on regularized denoising model of L1-L2 norm

电子行业常通过提取图像特征来对印刷电路板(Printed Circuit Board,PCB)进行缺陷识别。为了改善PCB图像的视觉效果,提升PCB无损检测的准确率,本文提出了一种基于L1-L2范数的正则项去噪模型的PCB图像去噪算法。首先采用非局部均值(Non Local Mean,NLM)滤波算法将提取的图像分解为结构和纹理两个部分,根据结构框架和纹理细节差异化的物理特性,分别使用Lasso回归算法和Ridge回归算法进行图像去噪,然后将Split Bregman迭代框架应用到去噪模型中,最后通过MATLAB软件平台对所提算法进行实验探究,并从视觉角度和去噪效果指标SNR、SSIM等多方面对算法进行评估。实验结果证明了基于L1-L2范数的正则项去噪模型的PCB图像去噪算法的有效性和可行性。

In the electronic industry,the defect recognition of printed circuit board(PCB)is often done by extracting image features.In order to improve the visual effect of PCB image and improve the accuracy of PCB nondestructive testing,this paper proposes a PCB image denoising algorithm based on L1-L2 norm regular term denoising model.First of all,NLM(Non Local Mean)filtering algorithm is used to decompose the extracted image into two parts:structure and texture.According to the physical characteristics of structure framework and texture detail differentiation,Lasso regression algorithm and ridge regression algorithm are used to denoise the image respectively,and then the split Bregman iterative framework is applied to the denoising model.Finally,the proposed algorithm was tested by MATLAB software platform Explore and evaluate the algorithm from multiple aspects such as visual angle and denoising effect indicators SNR,SSIM.The experimental results strongly prove the effectiveness and feasibility of PCB image denoising algorithm which is based on L1-L2 norm regular term denoising model.