基于智能激光视觉的犯罪现场特征提取
|
摘要
传统采用结构光视觉三维点云逐层三角网格化方法提取犯罪现场特征时,易受到噪声的干扰,导致特征提取准确率较低。提出基于智能激光视觉的犯罪现场特征提取方法,采用基于智能激光视觉检测方法检测犯罪现场指纹信息,规格化犯罪现场指纹图像,并获取指纹方向信息、指纹频率信息,对指纹图像进行滤波,通过向量点模式匹配算法提取滤波后的指纹图像的指纹特征。实验结果表明,所提方法可有效提取犯罪现场指纹特征,且对不同质量的指纹图像特征提取识别的平均准确率高达96. 98%,提取准确率较高;提取速度增长率高达30%,平均提取耗时约为110 s,特征提取效率较高。
The traditional method of triangulation of three-dimensional point cloud by layer of structure light vision is easy to be disturbed by noise,resulting in lower accuracy of feature extraction. This paper proposes a crime scene feature extraction method based on intelligent laser vision,adopts intelligent laser vision detection method to detect the fingerprint information at the crime scene,normalizes the fingerprint image at the crime scene,and obtains the fingerprint direction information and fingerprint frequency information,and filters the fingerprint image. Fingerprint characteristics of filtered fingerprint images are extracted by vector point pattern matching algorithm. The experimental results show that the proposed method can effectively extract the fingerprint characteristics of crime scene,and the average accuracy of different quality fingerprint image extraction and identification is as high as 96. 98%,and the accuracy of extraction is high. The extraction speed growth rate is as high as 30%,the average extraction time is about 110 s,and the feature extraction efficiency is higher.
The traditional method of triangulation of three-dimensional point cloud by layer of structure light vision is easy to be disturbed by noise,resulting in lower accuracy of feature extraction. This paper proposes a crime scene feature extraction method based on intelligent laser vision,adopts intelligent laser vision detection method to detect the fingerprint information at the crime scene,normalizes the fingerprint image at the crime scene,and obtains the fingerprint direction information and fingerprint frequency information,and filters the fingerprint image. Fingerprint characteristics of filtered fingerprint images are extracted by vector point pattern matching algorithm. The experimental results show that the proposed method can effectively extract the fingerprint characteristics of crime scene,and the average accuracy of different quality fingerprint image extraction and identification is as high as 96. 98%,and the accuracy of extraction is high. The extraction speed growth rate is as high as 30%,the average extraction time is about 110 s,and the feature extraction efficiency is higher.
引文
[1]赵雅彬,宋丹,郭威.潜血手印增强显现方法研究进展[J].化学试剂,2016,38(8):746-750.
[2]陆军,张艺竞,王成成,等.相位偏移gamma矫正算法的结构光三维视觉测量技术[J].哈尔滨工业大学学报,2017,46(9):182-188.
[3]秦绪佳,陈楼衡,谭小俊,等.结构光视觉三维点云逐层三角网格化算法[J].计算机科学,2016,43(s2):383-387.
[4]刘烁,王帅,傅焕章,等.软件定义的犯罪现场分析过程及其知识自动化方案[J].模式识别与人工智能,2016,29(10):876-883.
[5]曾祥通,张玉珍,孙佳嵩,等.颜色对比度增强的红外与可见光图像融合方法[J].红外与激光工程,2015,44(4):1198-1202.
[6]张晶,郑燕,刘倩,等.飞秒激光与微型角膜刀制瓣LASIK术后视觉质量比较[J].中华实验眼科杂志,2016,34(6):527-533.
[7]段良君,祝连庆,娄小平,等.激光扫描测头视觉系统结构参数优化[J].应用光学,2016,37(2):308-314.
[8]王琪龙,李建勇,沈海阔.双目视觉-激光测距传感器目标跟踪系统[J].光学学报,2016(9):178-186.
[9]王亚丽,魏振忠,张广军,等.视觉引导激光跟踪测量系统的Cayley变换校准方法[J].红外与激光工程,2016,45(5):240-245.
[10]张宾,宿敬肖,张微微,等.基于激光视觉的智能识别苹果采摘机器人设计[J].农机化研究,2016,38(7):60-64.
[11]刘子腾,白瑞林,王秀平.基于激光视觉的角焊缝图像特征点提取[J].焊接学报,2016,37(2):89-93.
[12]苑晶,刘钢墩,孙沁璇.激光与单目视觉融合的移动机器人运动目标跟踪[J].控制理论与应用,2016,33(2):196-204.
[13]刘曼,谭海佩,张月.基于灰色关联分析的运动角度规范性智能判断[J].激光杂志,2017,38(8):160-163.
[14]朱建新,沈东羽,吴钪.基于激光点云的智能挖掘机目标识别[J].计算机工程,2017,43(1):297-302.
[15]张荣辉,李福樑,周喜,等.一种基于激光与视频信息时空数据融合的行人检测方法[J].交通运输系统工程与信息,2015,15(3):49-55.
[16]程江珂,王胜男.基于激光智能视觉的芒果等级判断[J].激光杂志,2017,38(6):47-50.
[17]张小鸣,王燕萍.相位式激光测距系统优化设计及仿真[J].计算机仿真,2016,33(2):41-46.
[18]FAN G U,CHEN H,YINSHUI H E,et al.Method for Extracting Multi-Pass Welding Groove Feature Point Based on Laser Vision[J].Journal of Shanghai Jiaotong University,2016,50:62-65.
[19]YOUNUS Z S,MOHAMAD D,SABA T,et al.Contentbased image retrieval using PSO and k-means clustering algorithm[J].Arabian Journal of Geosciences,2015,8(8):6211-6224.
[2]陆军,张艺竞,王成成,等.相位偏移gamma矫正算法的结构光三维视觉测量技术[J].哈尔滨工业大学学报,2017,46(9):182-188.
[3]秦绪佳,陈楼衡,谭小俊,等.结构光视觉三维点云逐层三角网格化算法[J].计算机科学,2016,43(s2):383-387.
[4]刘烁,王帅,傅焕章,等.软件定义的犯罪现场分析过程及其知识自动化方案[J].模式识别与人工智能,2016,29(10):876-883.
[5]曾祥通,张玉珍,孙佳嵩,等.颜色对比度增强的红外与可见光图像融合方法[J].红外与激光工程,2015,44(4):1198-1202.
[6]张晶,郑燕,刘倩,等.飞秒激光与微型角膜刀制瓣LASIK术后视觉质量比较[J].中华实验眼科杂志,2016,34(6):527-533.
[7]段良君,祝连庆,娄小平,等.激光扫描测头视觉系统结构参数优化[J].应用光学,2016,37(2):308-314.
[8]王琪龙,李建勇,沈海阔.双目视觉-激光测距传感器目标跟踪系统[J].光学学报,2016(9):178-186.
[9]王亚丽,魏振忠,张广军,等.视觉引导激光跟踪测量系统的Cayley变换校准方法[J].红外与激光工程,2016,45(5):240-245.
[10]张宾,宿敬肖,张微微,等.基于激光视觉的智能识别苹果采摘机器人设计[J].农机化研究,2016,38(7):60-64.
[11]刘子腾,白瑞林,王秀平.基于激光视觉的角焊缝图像特征点提取[J].焊接学报,2016,37(2):89-93.
[12]苑晶,刘钢墩,孙沁璇.激光与单目视觉融合的移动机器人运动目标跟踪[J].控制理论与应用,2016,33(2):196-204.
[13]刘曼,谭海佩,张月.基于灰色关联分析的运动角度规范性智能判断[J].激光杂志,2017,38(8):160-163.
[14]朱建新,沈东羽,吴钪.基于激光点云的智能挖掘机目标识别[J].计算机工程,2017,43(1):297-302.
[15]张荣辉,李福樑,周喜,等.一种基于激光与视频信息时空数据融合的行人检测方法[J].交通运输系统工程与信息,2015,15(3):49-55.
[16]程江珂,王胜男.基于激光智能视觉的芒果等级判断[J].激光杂志,2017,38(6):47-50.
[17]张小鸣,王燕萍.相位式激光测距系统优化设计及仿真[J].计算机仿真,2016,33(2):41-46.
[18]FAN G U,CHEN H,YINSHUI H E,et al.Method for Extracting Multi-Pass Welding Groove Feature Point Based on Laser Vision[J].Journal of Shanghai Jiaotong University,2016,50:62-65.
[19]YOUNUS Z S,MOHAMAD D,SABA T,et al.Contentbased image retrieval using PSO and k-means clustering algorithm[J].Arabian Journal of Geosciences,2015,8(8):6211-6224.