License plate tracking recognition method based on event triggering

Abstract

The invention discloses a license plate tracking recognition method based on event triggering. The method comprises the following steps of performing vehicle detection and license plate extraction ona trigger frame at the trigger moment to obtain vehicle coordinates or a license plate area, and performing target vehicle judgment; and starting video stream tracking under the condition that whetherthe license plate quality evaluation of the target vehicle exceeds a specified threshold, performing license plate region judgment on a target vehicle region in the tracked video stream, and finallyobtaining a finally identified license plate through identification voting. According to the invention, bidirectional simultaneous frame-by-frame decoding and vehicle license plate tracking identification are carried out by taking the trigger frame as a reference in the video stream, so that the length of the intercepted video stream is effectively reduced and the tracking identification efficiency is improved; according to the tracked license plate, the confidence degree and the voting mechanism are integrated, and the situation that the single-frame license plate is poor in quality in some scenes is avoided.

Description

technical field [0001] The invention relates to a technology in the field of intelligent traffic control, in particular to an event-triggered license plate tracking and recognition method. Background technique [0002] The scene of license plate recognition is mainly divided into static scene and dynamic scene. Static scenes are mainly seen in parking lot checkpoints, residential property registration, etc., and the license plate is clear and fixed in size, and the recognition is good; dynamic scenes are often captured by various events in the traffic field, such as running a red light, honking, and speed measurement. Due to blurring, occlusion, shadows, overexposure, etc. during the capture process, the quality of the license plate triggered by a single frame is low, which seriously affects the capture rate of the license plate and the recognition accuracy of the license plate. [0003] Existing license plate recognition methods based on video streams either have a large a...

AI Technical Summary

Problems solved by technology

[0004] The present invention aims at the existing trigger event-based license plate recognition method, which is insufficient due to the low capture rate and low recognition rate caused by blurring, occlusion, shadow, and overexposure, and proposes an event-based license plate tracking and recognition method. In the video stream, the trigger frame is used as the benchmark, and the triggering target vehicle is used as the tracking target. Two-way simultaneous frame-by-frame decoding, vehicle tracking and license plate detection and recognition can effectively reduce the length of the intercepted video stream and improve the efficiency of tracking and recognition; according to a series of candidate license plates, Confidence-weighted voting mechanism is used to avoid the poor quality of single-frame license plate in some scenes

Images