Feature Modeling And Detecting Of High Cirrus Cloud In Infrared Imaging

Object detection is widely used in infrared image processing,especially in the field of remote sensing satellite infrared imaging.Because of the high infrared radiation energy of high altitude cirrus clouds,some local interference with energy concentration often occurs in the detection of infrared dim and small targets.For example,high-altitude cirrus clouds and other interferences often appear in the military field when using infrared imaging of extraterrestrial satellites to warn missile attacks.These interferences are called false alarm sources.In this case,an effective method is needed to detect these false alarm sources to eliminate their interference.In addition,in the field of meteorological environmental monitoring,the research on detection methods of high-altitude cirrus clouds is also one of the hot issues.Aiming at the false alarm source of high-altitude cirrus cloud in the infrared imaging detection of extraterrestrial satellites,the imaging characteristics of cirrus cloud are modeled and analyzed,and the detection method based on multi-feature modeling and analysis is studied.The research contents include: research on preprocessing and feature extraction methods of high-altitude cirrus infrared image,feature modeling and analysis of high-altitude cirrus cloud,and design and implementation of detection algorithm based on multi-feature combination.The specific work is as follows:1 The composition and morphology of high-altitude cirrus cloud and its imaging characteristics in infrared band are studied.2 The related theories of image preprocessing are studied.The denoising methods and image contrast enhancement methods aiming at the morphological characteristics of high-altitude cirrus clouds are emphatically analyzed,and the related preprocessing methods are experimentally tested.3 Various feature extraction methods of high-altitude cirrus clouds are studied,and their energy characteristics,shape characteristics and corner feature points are modeled and analyzed,and feature extraction experiments are carried out.4 The detection effects of different features in the feature extraction methods of high-altitude cirrus cloud are analyzed and compared,and the specific flow of the algorithm is designed based on these feature extraction methods,including the relevant module design of the algorithm,the experimental implementation steps and the performance analysis of the experimental test.5 The code of all feature extraction and detection algorithms are designed and completed,including cirrus cloud denoising,contrast enhancement,various feature extraction methods,image foreground and background segmentation methods,and final target frame selection detection code.In this dissertation,two kinds of high-altitude cirrus detection algorithms based on energy,shape and corner feature points detection of cirrus clouds are proposed by analyzing the feature modeling of high-altitude cirrus clouds.Both methods can effectively detect the specific location information of cirrus clouds in high altitude cirrus infrared images,and the detection accuracy is more than 90%,the false detection rate and the missed detection rate are less than 10%,and the running time of the algorithm is at MS level.