基于智能识别技术的铁路安检辅助分析装置研究

针对铁路安检X光图像判图高度依赖人工,时有发生漏检的问题,提出一种基于智能识别技术的铁路安检辅助分析装置;通过采用视频图像接口的硬件设计,解决与安检仪的适配问题;通过采用跟踪进程、分析进程、推送进程的多进程思路设计,解决60 Hz刷新率下的高精度分析显示问题;跟踪进程采用关键点差分算法实现安检X光图像跟踪,分析进程采用改进的残差网络(ResNet)实现特征提取,采用CenterNet算法实现禁限物品检测,推送进程通过检测键盘中断实现报警图像推送;经实验测试,该装置禁限物品识别准确率达92%,显示帧率达60帧/秒,可适配主流品牌安检仪,满足铁路车站安检辅助分析需求。

Research on Railway Security Inspection Auxiliary Analysis Device Based on Intelligent Identification Technology

Aiming at the problem that the judgment of railway security inspection X-ray image is highly dependent on manual work and sometimes missed prohibited articles, a railway security inspection auxiliary analysis device based on intelligent recognition technology is proposed. Through the hardware design of video image interface, the problem of adaptation to security inspection instrument is solved. Through the multi process design of tracking process, analyzing process and pushing process, the problem of high-precision analysis and display under 60Hz refresh rate is solved. The tracking process uses the key point difference algorithm to track the security inspection X-ray image, the analysis process uses the improved residual network (ResNet) to extract the features, the Centernet algorithm to detect the prohibited articles, and the push process uses the detection keyboard interrupt to push the alarm image. The experimental test shows that the identification accuracy of prohibited items of the device is 92% and the display frame rate is 60fps. It can be adapted to mainstream brand security inspection instruments,meeting the needs of auxiliary analysis of railway station security inspection.