| In recent years,with the rapid development of high-speed railway in China,railway foreign matter detection plays a vital role in railway safety.Due to the wide range of high-speed railway security detection,the sudden and unpredictable characteristics of foreign objects around the high-speed railway and the influence of bad weather,it poses a great challenge to the security around the highspeed railway.Therefore,a systematic solution is proposed for the perimeter security detection of high-speed railway.At the same time,the combination of deep learning algorithm and computer vision provides a solid foundation for the research of target detection algorithm.Based on the railway perimeter security project of Chinese Academy of Sciences and image processing,this paper mainly studies the denoising algorithm,target tracking and visualization scheme in the process of high-speed railway perimeter security detection.In terms of image enhancement,target detection and system implementation,the following research results have been achieved:(1)An image enhancement algorithm based on improved dark channel is proposed.This algorithm adopts the method of atmospheric model parameter extraction and histogram equalization to enhance the detection image.The algorithm not only improves the clarity of the detected image,but also enhances the robustness of deep learning target detection,and provides a more reliable input sample for the deep learning algorithm.(2)A perimeter security target detection method based on pedestrian re recognition is proposed.Based on Kalman filter and Hungarian algorithm,the target detection and quantity statistics of specific areas in the image are realized,and better detection accuracy and real-time are achieved.(3)Specific implementation of high-speed railway perimeter security system software.This topic provides a visual software for the perimeter security system of high-speed railway.The software integrates the functions of data transmission,equipment control,embedded development of detection algorithm and display of detection results,and provides a research and detection platform for perimeter security detection of high-speed railway. |
