Traffic flow monitoring method based on unmanned aerial vehicle, intelligent system and data set

Abstract

The invention discloses a traffic flow monitoring method based on an unmanned aerial vehicle, an intelligent system and a data set. The method comprises the steps of acquiring road traffic video dataof a traffic monitoring road segment, marking the acquired high-definition video data, establishing a training and testing data set based on a deep neural network; performing parameter learning and weight adjusting on the network by a multi-scale neural network according to the training and testing data set, performing positioning for obtaining the positions of vehicles in different types in a traffic video; processing a vehicle detecting result by means of a multi-characteristic fusion algorithm and a Bayes partile filtering algorithm, and furthermore tracking the vehicles in different typesfor obtaining a real-time flow monitoring result of the vehicles in different types. According to the method and the system, through performing accurate tracking on different detected vehicles in a low-altitude remote sensing video, real-time statistics is performed on the numbers of the vehicles in different types in the road traffic flow, thereby quickly and accurately monitoring the road traffic flow.

Description

technical field [0001] The invention relates to the technical field of UAV-based traffic flow monitoring in cities, in particular to a UAV-based traffic flow monitoring method, an intelligent system and a data set. Background technique [0002] Urban traffic flow monitoring plays a decisive role in the development and management of road traffic. However, currently widely used monitoring methods such as road cameras, induction coils, infrared rays, ultrasonic waves, etc., generally have single monitoring parameters, small coverage, and are susceptible to occlusion interference. The limitations of reliability and reliability have had many adverse effects on the accuracy and functional diversity of the traffic flow monitoring system; in recent years, with the rapid development of UAV and sensor integration technology, the collection of traffic information has also expanded from the traditional ground platform On low-altitude platforms, UAVs equipped with high-definition cameras...

AI Technical Summary

Problems solved by technology

[0002] Urban traffic flow monitoring plays a decisive role in the development and management of road traffic. However, currently widely used monitoring methods such as road cameras, induction coils, infrared rays, ultrasonic waves, etc., generally have single monitoring parameters, small coverage, and are susceptible to occlusion interference. The limitations of reliability and reliability have had many adverse effects on the accuracy and functional diversity of the traffic flow monitoring system; in recent years, with the rapid development of UAV and sensor integration technology, the collection of traffic information has also expanded from the traditional ground platform On low-altitude platforms, UAVs equipped with high-definition cameras have brought new data collection methods to the field of traffic monitoring, and can solve the shortcomings of traditional monitoring methods such as fixed positions, single monitoring methods, and low data quality; however, using UAVs to shoot The low-altitude remote sensing video contains a wealth of road traffic information and a variety of background noise, which also poses new challenges to traditional traffic flow analysis methods; therefore, providing new image and video signal processing methods for low-altitude remote sensing The development and improvement of the monitoring system is of great significance

Images