基于SURF特征的枪弹痕迹匹配方法

Bullet scrapping matching method based on SURF feature

为实现枪弹痕迹的自动比对与识别,提出将特征识别加速鲁棒特征(SURF)算法引入到弹壳痕迹匹配研究中,利用该算法提取弹底窝痕三维表面形貌特征,并采用随机抽样一致性(RANSAC)算法实现匹配优化。重点讨论了SURF特征点检测中参数调整及匹配效果关系,并借助美国国家标准与技术研究院(NIST)提供的弹底窝痕测试样本实现了最佳参数及识别条件的认定。实验结果表明,SURF算法对弹底窝痕表面形貌特征的提取与描述优异,在测试样本上可达到90 %以上的匹配率。

In order to achieve automated identification of the ballistics signatures,speeded-up robust features(SURF)and related algorithms are introduced into the feature matching implementation on cartridge case impressions.SURF is applied to extract three-dimensional(3D)topography features of breech face impression, and random sample consensus(RANSAC)algorithm is adopted to complete further matching optimization.The relationship between parameter adjustment in SURF keypoints detection and final matching result is mainly discussed,and the optimal parameters and identification conditions are verified with the help of the test samples provided by the National Institute of Standards and Technology(NIST)of USA.The experimental results strongly support the effectiveness of SURF algorithm on feature extraction and description of 3D breech face impressions, and indicate a matching rate of more than 90 % on the test samples.