Comprehensive monitoring method for regional road traffic system in cross-scale aerial platform

Abstract

The invention discloses a comprehensive monitoring method for a regional road traffic system in a cross-scale aerial platform and belongs to the field of aviation monitoring. The method specifically comprises steps that firstly, a semantic segmentation model and an object detection model are trained based on the convolutional neural network; images are acquired by an airship in real time in the monitoring area, a video after frame extracting is sent to the semantic segmentation model frame by frame for processing, and abnormal regions existing in each frame image are manually distinguished andlabeled; secondly, for a current frame image, the labeled adjacent abnormal regions are merged into a whole piece detection region, and a drone is assigned to the detection area; each drone goes to adesignated detection area, and the object detection model is utilized to perform detailed detection of low-altitude monitoring; and lastly, the abnormal position is determined, messages are uploadedand returned back to the airship, and corresponding warning or processing is performed. The method is advantaged in that through a cooperative work mode of the airship and the drone, the monitoring area is monitored roughly and in details, and complete coverage of the monitoring area is realized through relatively small cost.

Description

technical field [0001] The invention belongs to the field of aviation surveillance, and in particular relates to a method for comprehensive surveillance of a cross-scale space-based platform regional road traffic system. Background technique [0002] Road monitoring is mainly based on fast dome monitoring. The monitoring points are distributed at road intersections and key road sections where traffic and people flow are relatively concentrated. Regional road conditions in order to adjust the flow of vehicles at each intersection to ensure smooth traffic. In addition, it can detect and arrange to deal with road traffic accidents in a timely manner for monitoring violations of road vehicles, provide technical support for the detection of various cases such as traffic and public security, and greatly improve the level and efficiency of public security organs in law enforcement and case handling. [0003] Traditional road traffic monitoring is mostly roadbed monitoring, by plac...

AI Technical Summary

Problems solved by technology

[0005] However, the space-based platform surveillance methods for regional road traffic systems face many challenges: First, most of the space-based platform surveillance methods use unmanned aerial vehicles or airships with cameras, and the shooting distance is generally long. Vehicles and pedestrians in the video Targets are relatively small, placing high demands on the accuracy of surveillance methods

Images