Resumen
This paper describes a method that precisely estimates the position of images of traffic surveillance camera objects. We suggest a projection method with multiple traffic surveillance cameras through a local coordinate system into a global coordinate system. The transformation of coordinates uses detected objects, parameters of the camera and the geometric information of high- definition (HD) maps. Traffic surveillance cameras that pursue traffic safety and convenience use various sensors to generate traffic information. We suggest a transformation method with images of the camera and HD maps and an evaluation method. Therefore, it is necessary to improve the sensor-related technology to increase the efficiency and reliability of the traffic information. Recently, the role of the camera in collecting video information has become more important due to advances in artificial intelligence (AI) technology. The objects projected from the traffic surveillance camera domain to the HD domain are helpful to identify imperceptible zones, such as blind spots, on roads for autonomous driving assistance. In this study, we proposed to identify and track dynamic objects (vehicles, pedestrian, etc.) with traffic surveillance cameras, and to analyze and provide information about them in various environments. To this end, we conducted the identification of dynamic objects using the Yolov4 and DeepSort algorithms, established real-time multi-user support servers based on Kafka, defined transformation matrices between images and spatial coordinate systems, and implemented map-based dynamic object visualization. In addition, a positional consistency evaluation was performed to confirm its usefulness. Through the proposed scheme, we confirmed that multiple traffic surveillance cameras can serve as important sensors to provide relevant information by analyzing road conditions in real-time in terms of road infrastructure beyond a simple monitoring role.