去除红外图像条带噪声改进算法研究

由于红外焦平面阵列中各探测单元的响应特性不完全一致,导致像元之间存在一定的非均匀性,使红外图像中出现深浅相间的条纹状噪声,这种噪声严重影响了红外图像的质量和解析度。针对红外图像条带噪声的产生机理,在传统距匹配法的基础上,文中从图像灰度均值和方差的角度入手,结合红外图像数据量小、图像内容丰富的特点,将分段多项式拟合和中值预处理的思想分别应用到红外图像条带噪声的消除算法中,得到了改进的多项式拟合滤波法和基于中值预处理的分段拟合滤波法。理论分析和实验结果表明,文中提出的两种算法具有原理简单、实用性强的特点,能够有效地消除红外图像的条带噪声,提高图像信噪比。

Improved Algorithm for Removing Stripe Noise of Infrared Images

Due to the inconsistent response characteristics of infrared focal plane array detection units,stripping noise present in the infrared images affects the infrared image seriously both in quality and resolution.Based on the mechanism for the stripe noise of infrared image and the traditional moment matching method,this paper makes use of the angle of image grayscale average and variance and the characteristics of less data and rich content of the inferred images,and introduces the piecewise polynomial fitting and median pretreatment to the destripping algorithm.Theoretical analysis and experimental results show that the improved polynomial fitting filtering method and piecewise fitting filtering method based on median pretreatment are simple,practical,and can effectively eliminate stripe noise in infrared images to improve the image signal-to-noise ratio.