A weather phenomenon recognition system and method for assisting automatic driving of vehicles

Abstract

The invention discloses a weather phenomenon recognition system for assisting automatic driving of vehicles, which includes an ARM processor, and a vehicle-mounted camera, a remote server, a vehicle-mounted LCD display screen and a communication module respectively connected to the ARM processor; the vehicle-mounted camera shoots the vehicle in real time The weather image of the scene; the ARM processor divides the weather image into two parts: the sky image containing the weather information of the sky area and the ground image including the weather information of the ground area; the remote server will receive the weather image, the sky image and the ground image Import the weather phenomenon recognition model to identify the weather phenomenon type corresponding to the current weather image. The invention extracts the local features of the sky, the local features of the ground and the global features respectively, and imports the classifier after fusion to identify six weather phenomena of sunny, cloudy, snowy, rainy, foggy and sandy days, reducing the dependence on the number of samples , high-accuracy weather phenomenon recognition results can be achieved based on medium-sized weather image datasets.

Description

technical field [0001] The present invention relates to the technical field of image recognition, in particular to a weather phenomenon recognition system and method for assisting automatic driving of vehicles. Background technique [0002] Severe weather conditions can have a great impact on car driving. Severe weather conditions such as rain, snow, fog, sand and dust will reduce visibility and road friction coefficient, which may cause traffic jams and serious traffic accidents, with great potential danger. A weather phenomenon recognition system for assisting automatic driving of automobiles can effectively avoid serious traffic accidents and improve driving efficiency under severe weather conditions through real-time monitoring of the weather environment and comprehensive utilization of traffic information. For example, you can set the speed limit in bad weather conditions, remind you to keep the distance between cars, automatically turn on the wipers in rainy days, and...

AI Technical Summary

Problems solved by technology

However, these deep learning methods usually require large-scale data sets as support, and can only be effectively trained on high-end GPUs, which makes them very expensive in weather condition recognition. In addition, due to the complex background information in weather images , the current CNN model has certain defects in the discrimination of abstract weather information

Therefore, it is currently difficult to widely apply these methods to end-devices in the transportation field

Images