一种基于计算机视觉的胸环靶弹孔检测方法  

Bullet Hole Detection Method of Chest Ring Target Based on Computer Vision Technology

作　　者：黄应清 陈晓明 谢志宏 田钦文 HUANG Yingqing;CHEN Xiaoming;XIE Zhihong;TIAN Qinwen(Army Armored Force Academy, Beijing 100072, China)

机构地区：[1]陆军装甲兵学院,北京100072

出　　处：《兵器装备工程学报》2021年第11期157-162,221,共7页Journal of Ordnance Equipment Engineering

基　　金：军队重点项目(LJ20191A030128)。

摘　　要：提出了一种基于计算机视觉技术的弹孔检测方法。将传统的图像处理方法与深度学习的方法相结合,通过预处理、ROI检测和弹孔识别3个步骤实现胸环靶图像中弹孔的检测。实验结果表明,算法的精度较高,普适性强,实现了在不同靶场环境中胸环靶图弹孔快速、精准的检测。该算法性能良好,有效推动了基于计算机视觉的自动报靶系统的应用进程。A bullet hole detection method based on computer vision technology was proposed.Combining the traditional image processing method with the deep learning method,the bullet hole detection in the chest ring target image was realized through three steps:preprocessing,ROI detection and bullet hole recognition.The experimental results show that the algorithm has high accuracy,strong universality and fast detection speed,and completes the rapid and accurate detection of bullet holes in chest ring target images collected in different shooting ranges.The performance of the algorithm can meet the actual needs of the army shooting training,and effectively promote the application process of automatic target reporting system based on computer vision.

关 键 词：自动报靶系统 计算机视觉 深度学习 胸环靶 弹孔检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]