基于组合滤波的CMOS图像固定模式噪声降噪算法  

Fixed Pattern Noise Removal Algorithm for CMOS Image Based on Combined Filter

作　　者：管继祥 李林 张超 刘光东 罗文婷 Guan Jixiang;Li Lin;Zhang Chao;Liu Guangdong;Luo Wenting(College of Transportation and Civil Engineering,Fujian Agriculture and Forestry University,Fuzhou,Fujian 350100,China;Fujian Provincial Expressway Technology Consulting Co.,Ltd.,Fuzhou,Fujian 350001,China;College of Transportation Engineering,Nanjing Tech University,Nanjing,Jiangsu 211816,China)

机构地区：[1]福建农林大学交通与土木工程学院,福建福州350100 [2]福建省高速技术咨询有限责任公司,福建福州350001 [3]南京工业大学交通运输工程学院,江苏南京211816

出　　处：《应用激光》2022年第4期173-180,共8页Applied Laser

基　　金：国家重点研发计划项目(2021YFB3202901);福建省高校产学研合作重大项目(2020H6009);福建省科技厅产学研项目(KJB20005A);福建省交通运输厅项目(201912,KH200091A)。

摘　　要：固定模式噪声(fixed pattern noise, FPN)是CMOS(互补金属氧化物半导体)图像传感器中一种常见的噪声,该噪声会对相机成像质量产生很大的影响。为提高道路病害检测图像的质量,提出基于组合滤波的CMOS图像传感器固定模式噪声去除算法。首先使用均值滤波提取出CMOS传感器相机固定模式噪声的空间分布特征。随后使用提出的线性滤波方式去除绝大部分固定模式噪声。最后使用非线性滤波方式去除椒盐噪声。测试结果表明,经过该算法处理后,3组测试图像的峰值信噪比(PSNR)在28.75~32.50之间,图像均方差由处理之前的9.66~11.28降至处理之后的5.45~7.59,处理后的图像均方差相比处理之前降低了27.68%~43.61%不等。该算法有效地降低CMOS传感器相机中固定模式噪声对图像质量的影响,同时较好地保留道路病害图像的主要纹理特征,并且相对于传统的对CMOS传感器相机标定方法具有更高的实用性。Fixed pattern noise(FPN) is a kind of common noise in complementary metal oxide semiconductor(CMOS) camera sensors, which has an adverse impact on the image quality of the camera. In order to improve the quality of road disease detection images, this paper proposes a fixed-mode noise removal algorithm of CMOS image sensor based on combined filtering. Firstly, the mean filtering is used to extract the spatial distribution characteristics of the fixed-mode noise of the CMOS sensor camera. Secondly, the proposed linear filtering method is used to remove most of the fixed mode noise. Finally, nonlinear filtering is used to remove salt and pepper noise. Experimental results show that the peak signal-to-noise ratio(PSNR) of the three groups of test images is between 28.75 and 32.50, and the mean square error of the images decreases from 9.66 to 11.28 before processing to 5.45 to 7.59 after processing. The mean square error of the processed images was reduced by 27.68% to 43.61%. The proposed algorithm can effectively reduce the influence of fixed pattern noise in CMOS sensor camera on image quality and retain the main texture features of road damage images, which has higher practicability compared with the traditional calibration method for CMOS sensor camera.

关 键 词：激光成像 固定模式噪声 CMOS图像传感器 线性滤波 非线性滤波 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP391.4[自动化与计算机技术—控制科学与工程]