Research On Automatic Target-scoring System Based On FPGA

Manual target-scoring system is the main target-record method in the army's live-fire training.However,manual method has some defects in security and efficiency.The development of an automatic target reporting system with high accuracy,high speed and low cost is of great significance to effectively improve the level of military shooting training.This paper designs an automatic target-scoring system based on FPGA.The system includes image acquisition,image processing and data display.The image acquisition module uses an infrared camera to obtain the target image.The image processing module can accurately identify bullet holes and score a target.The data display module realizes the human-computer interaction of the location of point of bullet contact,current score and total score.The main work of this paper is as follows:Target image localization.Based on Ada Boost algorithm,the training is carried out in the computer by offline training.At the same time,according to the hardware characteristics of FPGA,the real-time detection of target image is completed by pipeline.Bullet hole recognition.In this paper,infrared image combined with frame difference method is used to detect bullet hole.The region of interest of bullet hole detected by frame difference method is identified by PCA algorithm.The nearest neighbor classifier is used to classify the region of interest,and the bullet hole region is retained.Finally,the centroid coordinates of the bullet hole are calculated by the ratio of the first moment and the zero moment to complete the location of the point of bullet contact.Target scoring calculation.Firstly,the circular Hough transform is used to identify the target center and the ring radius.Then,according to the position of the bullet hole and the effective area of the target,the bullet can be judged whether it hits the target or not.Finally,the score is judged by the distance between the bullet hole and the target center.By analyzing the morphological characteristics of target images with different radii,an improved circular Hough transform algorithm is proposed,which divides the image into blocks and detects circles locally.The algorithm greatly reduces the computational complexity and meets the storage requirements of the system.The automatic target-scoring system designed in this paper combines with the hardware platform,improves the image processing algorithm,can accurately obtain the bullet hole position,and calculate the score quickly.It can be seen from the experimental results that the bullet hole detection rate reaches 99.5%,the accuracy of score calculation reached 98.5%,and the processing speed reaches 50 ms / frame,which fully meets the requirements of actual shooting training.