A Video Stream Based Marksmanship System

With development of computer science and technology, computer is widely used in the field of automatic marksmanship training. But existing marksmanship systems are expected to improve their accuracy and real-time performance. In this paper, a video stream based marksmanship system is designed. Bullet holes can be accurately detected in real-time with simple hardware equipment and software system. Full automation, high accuracy and real-time is important properties of the system.The marksmanship system is composed of image sensor, embedded processor and computer. MT9P031 is selected as image sensor and STM32F407 is selected as embedded processor. Image sensor is responsible for video frame acquisition; embedded processor is responsible for some preparations and bullet holes detection in real-time; and computer is mainly used for providing graphical user interface, data processing and target reporting. There are several difficulties to realize the system: target information extraction, trapezoidal distortion correction and bullet holes detection in real-time. Target information, such as target position, ring spacing and feature point of the target, should be exactly extracted before bullet holes detection. Because of insufficient or excessive illumination, the difficulty of target information extraction is raised. Trapezoidal distortion is produced in the image because the camera cannot be placed fully opposite to the target in practice. There is a need to take an advance correction for the distortion to avoid the influence to next processes of target information extraction and bullet holes detection. Real-time detection of bullet holes is the most important stage in the marksmanship system. It’s difficult to detect the accurate location of bullet holes. Algorithms of bullet holes detection in real-time are studied in this paper according to practical requirement and solutions are proposed to solve these problems and some special cases. The experimental results are showed that the system can realize the real-time detection of the holes, and the accuracy of system is also guaranteed.