A method for predicting vehicle queuing dissipation time based on image self-learning

Abstract

The present invention relates to a method for predicting vehicle queuing dissipation time based on image self-learning, comprising: S1, preset the analysis area of ​​each lane of the entrance; S2, collect the image of the entrance; S3, segment the lane image from the image of the entrance; S4 1. The self-learning neural network analyzes the lane image to obtain the predicted time of vehicle queuing and dissipation; S5. Obtain the actual time of vehicle queuing and dispersal; S6. Calculate the predicted difference between the predicted time of queuing and disperse and the actual time; S7. Judge the predicted difference Whether they are all less than the first preset threshold, if it is judged to be yes, start to predict the vehicle queuing dissipation time in the next cycle, otherwise execute step S8; S8, add the lane image and the actual time of queuing dissipation to the sample relationship database, retrain and Update the self-learning neural network of step S4. Compared with the prior art, the present invention realizes prediction model evolution through image self-learning, and can accurately predict vehicle queuing dissipation time.

Description

technical field [0001] The invention relates to the technical field of intelligent transportation, in particular to a method for predicting vehicle queuing dissipation time based on image self-learning. Background technique [0002] The purpose of signal dynamic timing is to balance the flow ratio of the entrance lanes in all directions, so as to ensure that the vehicles waiting in line will be released in one green light cycle. Therefore, it is necessary to accurately predict the vehicle queue dissipation time to ensure the effectiveness of signal control. To improve the efficiency of intersection traffic, the prediction of vehicle queuing dissipation time is usually based on the signal perception of the intersection, and the key lies in the perception of the queuing status of each entrance. [0003] The traditional intersection signal perception mainly uses electromagnetic induction coils. Since the coils need to cut the road surface, rainwater penetration and vehicle roll...

AI Technical Summary

Problems solved by technology

[0003] The traditional intersection signal perception mainly uses electromagnetic induction coils. Since the coils need to cut the road surface, rainwater penetration and vehicle rolling will easily cause damage to the coils. Therefore, the method of using electromagnetic induction coils to sense intersection signals has poor reliability, The disadvantage of being difficult to repair, in addition, the coil induction is essentially a point-type sensing, its sensing range is limited, and it is impossible to obtain all signals at the intersection

Images