Video Tracking Methods And Its Experimental Research For Air-to-surface Target In Complex Background

Air-to-surface target tracking in complex backgrounds is one of the keytechnologies of the airborne photoelectric systems. Because of the complexitiesand the differences of backgrounds and the lighting conditions changes, as wellas the mobility, stealth, low signal-to-noise ratio, obscure or discontinuity of themoving target, it is usually difficult to achieve sustained, steady and reliabletarget tracking in airborne complex backgrounds.Both Visible and infrared sensors have their own advantages anddisadvantages. The ground target has more details but its main body may losssome information in the visible image. The main body of target becomes moreclearly in infrared imaging system, but its details are not as good as visible lightimaging results. This paper presents target tracking methods in complexbackgrounds both in visible video tracking and image fusion tracking with thevisible and infrared sensors configured in airborne photoelectric pod and theirrespective advantages and disadvantages.The mainly work and innovations of this paper:(1)In the case of visible light sensor, this paper improves the classicCamshift tracking algorithm. The original algorithm uses only a single colorcharacteristic. The improved algorithm adds texture and movementcharacteristics. Additionally, the algorithm also introduces the Kalman filter. Thepoint tracking and region tracking are combined to improve the trackingperformance.(2)Since the infrared sensor and visible light sensor have their respective advantages and disadvantages, multi-sensor fusion tracking can also be used toimprove the tracking performance under complex backgrounds. This paperproposes respective tracking framework of the pixel level, characteristic leveland decision level image fusion.(3)Because it’s difficult to obtain actual airborne images, an airborneanalog observation system and its supporting software is designed andimplemented in the laboratory environment. The system is used to simulate theaircraft earth observation images and validate the mentioned algorithms in thispaper.