A Vehicle License Plate Recognition System Based on Spatial/Frequency Domain Filtering and Neural Networks

License plate recognition based on temporal redundancy

“…Although many works propose approaches to solve a single subtask at a time, there are works proposing techniques to perform the entire ALPR pipeline [Donoser et al, 2007;Guo and Liu, 2008;Kocer and Cevik, 2011;Ozbay and Ercelebi, 2005;Qadri and Asif, 2009;Wang et al, 2010].…”

“…Furthermore, they also combine multiple detections in order to make the recognition robust to noises presented in a single frame. The work proposed by Wang et al [2010] proposed a technique able to locate the license plate using horizontal scans of contrast changes, segment the plate using lateral histogram analysis and recognize the characters using an Artificial Neural Network for Italian license plates. Kocer and Cevik [2011] proposed a work to locate the region of the image with the most transition points assuming that it corresponds to the license plate.…”

An Automated Vehicle License Plate Recognition System

“…5 Mu-Liang Wang and colleagues used horizontal scans of repeating contrast changes for plate recognition, but it suffers from a ringing effect that occurs along the edges of the filtered spatial domain image. 6 Fully connected feed forward artificial neural networks with sigmoidal activation functions have also been used for character recognition, but the successful number plate identification rate is only 80 percent and processing time is 15 seconds. 6 One study proposed a license plate detection method based on sliding concentric windows and a histogram, but it was both time-consuming and only suitable for Taiwanese plates.…”

“…6 Fully connected feed forward artificial neural networks with sigmoidal activation functions have also been used for character recognition, but the successful number plate identification rate is only 80 percent and processing time is 15 seconds. 6 One study proposed a license plate detection method based on sliding concentric windows and a histogram, but it was both time-consuming and only suitable for Taiwanese plates. 7 Zhen-Xue Chen and colleagues combined the rectangle shape, texture, and color features to extract the license plate, 8 and had a success rate of 97.3 percent, but the process was too computationally complex and time-consuming.…”

Identificação De Caracteres Para Reconhecimento Automático De Placas Veiculares

“…Selecionamos o trabalho que apresenta o melhor resultado geral, e o que apresenta o terceiro melhor resultado, uma vez que o segundo melhor aborda o mesmo problema que o primeiro. (Lee et al, 2004) Template Matching 95.7% Taiwan (Duan et al, 2004) Hidden Markov Model 97.5% Vietnam (Shi et al, 2005) Neural Network 89.1% Grécia (Chang et al, 2004) -94.2% Multi nacional (Deb e Jo, 2009) Self Organizing OCR 95.6% Taiwan (Wang et al, 2010) Neural Network 98% Itália (Kim et al, 2000) SVM 97.2% Coréia (Comelli et al, 1995) Template Matching 98.6% Itália (Capar e Gokmen, 2006) Neural Network 97.7% China Tabela 3.1: Resumo do estado da arte de reconhecimento de caracteres…”