Target Recognition And Tracking In The Test Of Rocket Sled Ejection

Ejection seat is an important life-saving equipment. The criteria of ejection seat have been raised by modern warfare featuring in high speed and mobility. Rocket sled is a kind of rail vehicle propelled by rocket engine which is mainly used in ejection seat test. To simulate the flying environment, rocket sled is capable of twice the speed of sound in the test, in which process pilot ejects. In current test, the whole process is recorded by high-speed camera manually. Affected by sonic boom and reaction speed, target might get lost in recording process. To solve that problem, this paper launches researches on auto track recording system from the aspects of algorithm and hardware.The whole test process can be divided into four parts:sled acceleration, pilot ejection, seat separation, steady descent. Several methods were implemented to track the rocket sled and detect the pilot. Camshift algorithm is an essential tracking algorithm in this paper.The background wall was designed according to the real proving ground. To simulate the moving rocket, a reflector was used to project the rocket sled image to the background wall. Driven by a small turntable, the reflector was able to spin at different speed, thereby simulates the moving rocket. In the experiment, a web camera was mounted on tracking turntable, images captured by the web camera were sent to computer. After the processing based on OpenCV, an offset was calculated and sent to STM32.Then STM32sent out pulses at different frequencies to control the turntable to track the target.