非制式枪射击弹头痕迹自动识别系统及性能测试  

Introduction to the Automatic Identification System of Non-standard Firearms' Bullet Marks and Its Performance

作　　者：周志飞[1] 李轶昳[1] 鲍立垠[1] 赵衍运[2] 宋荣健 王晓琳[1] 

机构地区：[1]公安部物证鉴定中心,北京100038 [2]北京邮电大学,北京100876

出　　处：《刑事技术》2016年第2期99-102,共4页Forensic Science and Technology

基　　金：公安部重点研究计划项目(No.201202ZDYJ006)

摘　　要：本文对非制式枪射击弹头痕迹自动识别系统进行了详细介绍,并对该系统的自动识别性能进行了测试和评估。该系统利用铝箔模片和弹头痕迹展平装置采集非制式枪射击弹头痕迹图像,对采集到的图像进行特征提取,并存储于数据库中,系统能够以较高的准确率和检索速度完成非制式枪射击弹痕的自动比对识别任务。该系统提供了多种基于基础图像处理操作的专家比对模式,开发了弹头痕迹图像的重叠比对和分割比对方式。经测试,系统的自动识别结果在测试库中排名第一的比例为64.5%,排名前30的比例为90.3%。该系统的投入应用可节省人力物力,提高射击弹头痕迹同一认定的检验效率,为案件的快速串并提供有力的技术支持。The article reports and evaluates the automatic identification system of non-standard firearms＇ bullet marks. The system captures images of bullet marks through the bullet-nose-cone-rifling-trace-developing device and adhesive aluminumfoiled tape, standardizes and extracts the characteristics of the captured images, and stores the standardized images into its database. Based on the automatic comparison algorithms of good performance, the system can complete a speedy identification with high accuracy. The images, most similar with those of the goals, can be retrieved from the database so that the cartridge examiners could make further decision upon the retrieved images. In addition, the system can also provide a variety of basic image processing operations for artificial contrast mode, and conduct the comparison model of either overlapping or segmentation based on the image characteristics of non-standard bullet marks. Still, the system furnishes the bullet mark image comparison reports that generate automatically and show the basic information from the image management database. In brief, the system is of highly comprehensive automatic identification and management. For test samples, 736 distinct non-standard firearms were collected from different provinces and cities. Each firearm fired two bullets, one of which was imaged in the reference samples database, the other in the test sampled database. 185 samples were randomly selected from test samples database and then correlated against the reference samples database. Among the results of automatic identification from the test database are 64.5 % ranking at No. 1 and 90.3 % on the top 30, suggesting that Automatic Bullet marks Identification System for non-standard firearms has a certain practicality. The system could improve the efficiency for bullet marks identification and examination, save manpower and material resources in contrast to the manual checking way of one by one under microscope, thereby providing better support with linking a series of cases for

关 键 词：自动识别 射击弹头痕迹 非制式枪 

分 类 号：DF794.1[政治法律—诉讼法学]