Zeng Shengke, Amandus Harlan E, Amendola Alfred A, Newbraugh Bradley H, Cantis Douglas M, Weaver Darlene
Division of Safety Research, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV, USA.
J Transp Technol. 2014 Jul;4(3):216-255. doi: 10.4236/jtts.2014.43022.
The homicide rate of taxicab-industry is 20 times greater than that of all workers. A NIOSH study showed that cities with taxicab-security cameras experienced significant reduction in taxicab driver homicides.
Minimum technical requirements and a standard test protocol for taxicab-security cameras for effective taxicab-facial identification were determined. The study took more than 10,000 photographs of human-face charts in a simulated-taxicab with various photographic resolutions, dynamic ranges, lens-distortions, and motion-blurs in various light and cab-seat conditions. Thirteen volunteer photograph-evaluators evaluated these face photographs and voted for the minimum technical requirements for taxicab-security cameras.
Five worst-case scenario photographic image quality thresholds were suggested: the resolution of XGA-format, highlight-dynamic-range of 1 EV, twilight-dynamic-range of 3.3 EV, lens-distortion of 30%, and shutter-speed of 1/30 second.
These minimum requirements will help taxicab regulators and fleets to identify effective taxicab-security cameras, and help taxicab-security camera manufacturers to improve the camera facial identification capability.
出租车行业的凶杀率比所有行业工人的凶杀率高出20倍。美国国家职业安全与健康研究所的一项研究表明,安装了出租车安全摄像头的城市,出租车司机凶杀案显著减少。
确定了用于有效出租车面部识别的出租车安全摄像头的最低技术要求和标准测试协议。该研究在模拟出租车中,在各种光照和出租车座位条件下,以不同的摄影分辨率、动态范围、镜头畸变和运动模糊拍摄了10000多张人脸图表照片。13名志愿者照片评估员对这些面部照片进行了评估,并投票选出了出租车安全摄像头的最低技术要求。
提出了五个最坏情况场景下的摄影图像质量阈值:XGA格式分辨率、高光动态范围1 EV、黄昏动态范围3.3 EV、镜头畸变30%以及快门速度1/30秒。
这些最低要求将帮助出租车监管机构和车队识别有效的出租车安全摄像头,并帮助出租车安全摄像头制造商提高摄像头面部识别能力。