Infrared Small Target Detection Based On Image Fusion Technology

The capture and tracking of infrared small targets are widely used in the fields of infrared precise guidance,remote sensing,space exploration,aviation navigation and etc.However,the application environment sometimes forces us to keep certain distance between the imaging system and the targets,resulting smaller targets in the camera's field of view.Meanwhile,due to the jitter of the infrared imaging equipment,the noise of the imaging equipment and other factors,the weak target signal is submerged in the complex background.Besides electromagnetic signal that stealth fighter and etc radiate into the air is getting weaker and weaker.Radar and other explorational devices which use the echo of reflected target to search and locate have not met the demand of reality.These existing problems make the detection and tracking of weak small targets in infrared image sequences more difficult and challenging.Infrared thermal imaging technology,due to its good concealment performance,features of all-weather and day-night imaging,good mobility and convenient configuration,is very conducive to carrying on air early warning and unmanned aerial vehicles.If surveillance system can identify enemy targets in the early time,Then,the target is locked and tracked,thus,the detection of area target tracking and identification can be omitted and the response time of the own defense system can be greatly extended,or in the area target phase of the second detection of the target,reduce the false alarm probability of detection,and sends out the alarm to the defense system.Therefore,the design of a robust infrared small target detection algorithm has important realistic meaning and practical value.In this paper,two different small target detection algorithms are designed on the basis of predecessors' work: Small target detection based on quantum genetic algorithm with local contrast enhancement and small target detection of local intensity and gradient fusion under bilateral filtering.One is to use the quantum genetic algorithm to optimize and obtain structural elements,and construct corresponding structural elements for different background and other undulating edges to conduct morphological filtering,then energy of small targets are enhanced by local contrast,finally threshold separation technology is employed to extract small targets.The other detection algorithm chooses bilateral filtering technology in clutter suppression,aiming to consider spatial neighborhood information and gray similarity characteristics to achieve the purpose of noise canceling and de-noising and preserving edges.Then,from the perspective of human vision,an algorithm of local intensity and gradient fusion is proposed to enhance targets and suppress the background clutter.Finally,small targets are detected successfully by adaptive threshold separation technology.Experimental evaluation shows that the proposed two algorithms are effective and robust in background clutter suppression,especially the introduction of image fusion technology,which can effectively suppress Gaussian noise and undulating edges.Compared with the traditional detection algorithm,the detection performance is significantly improved and the robustness is better.The research work in the paper has certain theoretical basis and practical significance for image edge filtering and target enhancement under different backgrounds,and also has certain inspiration and practical value for realizing high detection probability and low false alarm probability of small targets in the field of infrared early warning.