Traffic incident detection device and method based on audio

Abstract

The invention discloses a traffic incident detection device and method based on an audio and relates to the technical field of intelligent transportation. The device comprises a microphone array module, a microphone signal receiving module, a noise reduction and dereverbration module and an incident detection module; a digital microphone array serves as sound pick-up equipment and is small in volume and low in cost, and the acquisition and recognition for incidents such as vehicle starting checking, driving passing, braking, crash, explosion and calling for help which occur on the road by the digital microphone array are achieved; an improved MVD algorithm is used for processing signals and capable of effectively restraining noise and reverberation at the same time, and has the advantages of being good in robustness, good in real-time property, high is stability and low in algorithm complexity; finally, through a Convolutional Neural Network, the probability of a traffic incident is obtained, and therefore the type of the traffic accident is determined, and the problem is improved that when a traditional neural network layer number deepens, a global optimal point is not easy to find out; for different traffic conditions, the traffic incident detection device and method based on the audio have the advantages of low cost, wide application range, all weather operation, and high detection precision.

Description

technical field [0001] The invention belongs to the technical field of intelligent transportation, and in particular relates to an audio-based traffic event detection device and method. Background technique [0002] With the rapid development of social economy, there are more and more vehicles on urban roads, national roads and expressways, resulting in increasingly severe traffic congestion, accidents, public security, etc., and frequent traffic delays. However, most traffic anomalies are caused by non-recurring traffic events. Therefore, the acquisition of information on traffic events, especially abnormal events such as congestion, accidents, public security, etc., is of great significance for reducing the impact and duration of traffic events and implementing reasonable road network optimization. [0003] At present, most detection methods of traffic incidents use video technology. The traffic information acquisition technology using video is to collect video informati...

AI Technical Summary

Problems solved by technology

In terms of data processing of audio-based traffic incident detection devices, today's mainstream methods are intelligent detection algorithms, such as artificial neural networks. This method has a high detection rate and a low false alarm rate, but it is difficult to find the overall optimal point, while degrading the recognition performance of traffic events

Images