A Research On Automatic Target Scoring System Based On MSER

In the military shooting training,the traditional manual reporting method is susceptible to the environment of the site and weather,whose efficiency is not good enough and there are security risks.In recent years,the researchers are seeking breakthroughs from scoring algorithms and equipment to overcome the shortcomings of traditional reporting systems,and using electronic aids to establish automated systems for shooting training.This paper studies the image processing algorithm which realize the automatic reporting function.The automatic scoring system in this paper includes a lower computer of ARM and a host computer.The lower computer uses a webcam to obtain the chest ring target images in the shooting training.Each camera corresponds to a ring target.In order to enable the system to report bullet holes from multiple targets at the same time,multi-thread technology is used to manage multiple cameras and real-time target reporting tasks to fully utilize the CPU resources of the host computer.Each thread realizes bullet hole recognition algorithm for the captured image,and finally displays information such as the new bullet holes position and the number of rings in the system interface.In the host computer,the bullet hole recognition algorithm based on Maximally Stable Extremal Regions(MSER)is studied.In order to adapt the MSER algorithm to the environment,the MSER algorithm is improved according to the integral image-based threshold adaptive algorithm,so that the MSER algorithm parameters can be dynamically adjusted(MSER-I).At the same time,in order to improve the computational efficiency of the MSER-I algorithm,the possible bullet hole regions are quickly determined by interframe difference,the single target detection strategy is proposed to simplify the algorithm flow of MSER.In the system,using the single target strategy in each possible bullet hole region,the MSER-I algorithm is executed in the region to obtain accurate candidate bullet hole regions.In addition,the image acquired by the camera is distorted,therefore the bullet hole pixels in the captured chest ring target are mapped into the standard chest ring target by perspective transformation,and the chest ring target area and the background area can be distinguished,thereby eliminating candidate bullet holes outside the chest ring target.For the candidate bullet holes in the chest ring target,the bullet hole feature is used to exclude the nonbullet hole area,and according to the bullet record,the bullet holes that already been reported on the ring target are excluded.The newly added bullet holes are finally identified through the procedure above.The experimental results show that the MSER-I algorithm proposed in this paper can obtain the bullet hole area more accurately in a complex environment.The parameter adaptive strategy can cope with the changing ambient illumination changes so that the detected bullet hole contours match the actual bullet holes as much as possible.In addition,interframe difference and the single target detection of MSER algorithms are used to improve the efficiency of bullet hole detection.In the bullet hole screening stage,it is possible to effectively extract new bullet holes based on the bullet hole overlapping rate and the feature of bullet hole.Comprehensively analyzed,the system meets the detecting accuracy and real-time requirements.