Feature Modeling And Detecting Of Snow Moutain False Alarm Sourece In Infrared Imaging

Infrared small target detection technology is widely used in military and commercial fields.At the same time,exisitence of high radiant energy false alarm sources causes interference to target detection,which leads to high false positive rate.As a high radiant energy object,snow mountain is a typical false alarm source in military early warning and interception system.It is a new way to improve small target detection accuracy by detecting infrared snow mountain and separating it from small infrared targets.This thesis mainly studies the characteristic modeling and detection of infrared snow mountain false alarm source.Energy and morphological characteristics of infrared snow mountain were extracted and described mathematically.By analyzing the characteristics of snow mountain and background,efficient and accurate methods for detecting snow mountain false alarm source was promoted.In this dissertation,the main reasearch contents can be summarized as follows:(1)In this thesis,infrared imaging principle and issue of image deterioration were analyzed.Image preprocessing methods were studied,which mainly includes various denoising algorithms and contrast enhancement algorithms.In order to obtain more accurate detection result,morphological processing methods for binary images were studied as well.(2)Based on high saliency from the high radiation property of snow mountain false alarm source in infrared images,several commonly used saliency characteristic models with high computational efficiency and simple principle were studid,and were applied to snow mountain false alarm source.The saliency characteristics in frequency domain of infrared snow mountain were analyzed and simulated emphatically in this thesis.(3)In this thesis,according to the texture characteristics of snow mountains,several texture features were studied,including local entropy,fractal dimension and grayscale cooccurence matrix(GLCM).Texture characteristics of snow mountain were obtained by analyzing.These features showed different performance,indicating that they can be applied in different scenarios.(4)According to researches above,two types of snow mountain false alarm source detection methods were proposed,based on saliency feature and texture feature.One is the use of local entropy texture feature,and the other is the use of fractal texture feature.Both methods can detect the snow mountain area well in different scenes.Furthermore,algorithms were impleted and verified on the Matlab platform.The imaging quality,accuracy and timeliness of the test results of the two methods were analyzed,which conforms to the indexes required by the project.