The Detection Of Infrared Small Target Based On Temporal And Spatial Domain Filter

In order to improve the capability of the infrared small target detection, the papertook geostationary orbit staring detectors as application background, and put forwardtwo small target detection algorithms, three-dimensional matched filtering and targetdetection based on projection algorithm. According to the intrinsic jitter characteristic ofmoving platform and the problem of low target association in image sequences capturedby the gaze detector, we proposed a new jitter compensation algorithm. The main worksof this paper is as follow:(1)By analyzing the infrared small target image characteristics, the algorithmsimulation input was given a theoretical basis;(2)According to the background characteristics of infrared cloud, we constructedbackground model based on gray using the principle of least squares method, so as forthe image sequences jitter estimating and the image sequences registration;(3)The principle of the maximum signal to noise ratio and Cauchy-Schwarzinequality was used for to deriving the general mathematical formula of the threedimensional matched filter. We gave the general process and conducted a simulationalgorithm of the three-dimensional matched filtering algorithm for the sequence ofimages, and effectively verified the three dimensional matched filtering algorithm forthe detection of small targets with low signal to noise ratio with superior performance,however, the three dimensional matched filter would behave poorly when velocitymismatched, which required more intensive velocity directional filter if you want it withsuperior performance computing, and all above put forward requirement for the storageand computing capacity of the existing computer.(4)For two-dimensional images obtained by the projection algorithm, we designeda non-related point removal algorithm associated with hierarchical theory of voting,which effectively removed the isolated points and weak related points, retaining thetarget trajectory, and we designed and calculated the classification parameters of themulti-target trajectory by introducing the principle of hash, effectively classifying themulti-target track, and for both algorithms the simulation is given.(5)In this paper, we designed the infrared small target detection software systembased on GUI using the research before, and the software can realize the functionmodule which contains the jitter sequence generation, jitter calculation for pendingregistration frame, three-dimensional matched filtering algorithm for small targetdetection and small target detection based on projection algorithm.The spatial filter based infrared small target detection studied in this paper can provide theoretical basis for target detection in orbit.