Research On Driving Intention Of Preceding Vehicle Based On Machine Vision

Statistics show that rear-end collision accounts for 30% of traffic accidents.Most of the rear-end collisions are caused by unclear judgments about the driving intentions of the preceding vehicles.In this paper,the machine vision method is used for vehicle detection and light recognition.The driving intention of the preceding vehicle is analyzed,which can effectively avoid the occurrence of rear-end collisions and provide a way to understand road information for autonomous driving.Aiming at the problem that taillight detection is greatly affected by the complicated traffic environment,a method of combining the vehicle detection with the taillight detection is proposed.The image acquisition device is installed in front of the vehicle to collect the image of the road ahead,and the vehicle image data set is made.The Faster R-CNN is constructed,and the parameters in the network model are set.The convolutional neural network is trained and tested.The Faster R-CNN is used to complete the vehicle detection in front of the current lane.In the RGB color space,the taillights are detected and identified by using the R channel threshold to detect and identify the taillights,which can cause it not accurately to detect red models.The accuracy of the method of detecting taillight with only position relation is not high.Aiming to improve the accuracy,an improved light language recognition algorithm based on histogram feature parameters is proposed.The taillight area is extracted in the HSV color space,and the taillight pair is detected by using three characteristic correlations of gradient histogram,color,and position.The taillight state is determined according to the histogram characteristic parameter of the tail light image.The algorithm proposed in the paper is verified by real road environment.Both vehicle detection and taillight status detection have higher accuracy.The method can overcome most of the interference factors such as illumination and complex traffic background.Therefore,a more accurate analysis of the driving intention of the preceding vehicles can be achieved.