Research On The Techniques Of Air Target Detection In IR Tracking System

Combined with the project of “XX IR tracking system”,this paper focus on the taskof air target detection,whish is the subtask of IR tacking subsystem of the project.Thistask demand detecting air target at varing distance.If the target is at a far distance, itsprojected image is very small and dim,in the other hand,if the target is at a neardistance,its projected image is larger.So we do the reseachring of small dim target andaera target detection.The dissertation is comprised of three parts:In the first part,the conventional IR dim small target detection algorithm isstudied.when the field of view contained ground background,which is so complex thatthe detection result contains some false alarms in the gound region.A method of skyregion extraction is proposed out to solve this problem,the experimental results showthat this method can extract the sky region correctly,so could exclude the false alarms inthe ground region.In the second part,the small target detection algorithm based on auto scale selectionis studied.In the real world,a incoming target changes its size in the image,but theconventional filtering metolds cannot adapted to this varing,this limits the performanceof detector.This paper studied the auto scale secletion method to extaction the target’ssize and position information,which can be used to solve the varing size problem. Theexperimental results show that this method can detect size varing target very well.In the third part,this paper focus on the detection of area target.Fistly,the targetdetection algorithm based on thresholding segmentation is studied.When thebackground is complex,these metholds cannot get satisfying segmentation results.Thispaper proposed a methold of target segmentation based on morphologicalgradient,which takes advantage of edge and amplitude information. The experimentalresults show that this method get better performance.