License plate recognition algorithm

Abstract

The invention relates to the technical field of license plate recognition, in particular to a license plate recognition algorithm, which comprises the following steps: S1, license plate positioning: 1) positioning a license plate by using an improved YOLOV3 detection network so as to obtain angular point coordinates of the license plate; 2) cutting the license plate, and sending the cutting result to a recognition network, and S2, recognizing the license plate: 1) generating sequence features: converting a license plate bounding box into a series of feature vectors; 2) sequence labeling: training an RNN model with GRU by using Adam and BP algorithms to label sequence features; and 3) sequence decoding: applying the CTC to an output layer of the RNN, analyzing a marking result of the RNN and generating a final recognition result, on the whole, the algorithm is finer in detection of small objects such as small license plates and more accurate in selection of anchor point frames, and meanwhile, the network can better utilize fine-grained features of different levels of low-level and high-level to realize multi-scale detection.

Description

technical field [0001] The invention relates to the technical field of license plate recognition, in particular to a license plate recognition algorithm. Background technique [0002] License plate recognition is one of the important components of modern intelligent transportation systems, and it is widely used. It is based on digital image processing, pattern recognition, computer vision and other technologies, and analyzes the vehicle images or video sequences captured by the camera. The unique license plate number of a car, so as to complete the identification process, through some follow-up processing methods, parking lot fee management, traffic flow control index measurement, vehicle positioning, car anti-theft, expressway speeding automatic supervision and other functions can be realized, which is very important for maintaining traffic safety. It is of practical significance to improve urban public security, prevent traffic jams, and realize traffic automation manageme...

AI Technical Summary

Problems solved by technology

[0003] The existing license plate recognition algorithm is not fine enough to detect small license plates, and the selection of the anchor frame is not accurate enough, so a license plate recognition algorithm is needed to improve the above problems

Images