摘要
前视红外图像质量对于目标检测与跟踪算法能否准确定位目标具有十分重要的作用。目前对自动目标识别领域前视红外图像质量的研究大都与场景内容相关,而忽视了成像降质效应对图像质量的影响。本文通过分析噪声和模糊两种降质效应的成因,结合前人的模拟方法,得到了高性能前视红外成像设备获取的图像中噪声和模糊模拟的新方法。将该方法用于分析不同程度噪声和模糊对图像质量的影响,结果显示,噪声对于图像质量有害,而模糊对于图像质量并不一定有害。基于这一结论,采用去噪方法对前视红外图像进行预处理。对比实验结果表明,该方法比去模糊方法能更有效地提高匹配概率。
Forward Looking Infrared （FLIR） image quality is of great importance for the detection and tracking algorithms to accurately locate the targets. In the field of Automatic Target Recognition （ATR）, present researches on FLIR image quality are mainly related to the content of the scene, while ignoring the influence of imaging degradations on image quality. Through analysis of the cause of noise and blur, and combination with the former simulation methods, a new method was established for noise and blur simulation in images acquired by high-performance FLIR equipment. This method has been applied to analyze the effect of different degrees of noise and blur on image quality, and it can be concluded that noise has adverse effect on image quality, while blur is not always deleterious. Based on this conclusion, denoising method has been adopted in the preprocessing of FLIR image. Comparison among experimental results reveals that denoising method can improve the matching probability more effectively than sharpening method.
出处
《光电工程》
CAS
CSCD
北大核心
2012年第8期55-62,共8页
Opto-Electronic Engineering
关键词
前视红外
降质效应
噪声
模糊
图像质量
forward looking infrared
degradations
noise
blur
image quality
