The application of a CICA Neural Network on Farsi license plates recognition

Akhtari, Mojdeh; Faez, Karim

doi:10.1109/his.2010.5600081

Cited by 1 publication

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…5 illustrates the schematic of this method. It should be noted that for pairs with similar classes like (2, 2) and pairs with symmetrical classes such as (3,2) and (2, 3), separate SVMs are not made. The author maintains that the presented method has a higher accuracy than the method used in [14].…”

Section: B the One-against-one Methodsmentioning

confidence: 98%

“…In the area of character recognition, neural networks are the most often used types of classifiers [3,9]. These networks are capable of separating several classes at the same time and can also separate the classes nonlinearly.…”

mentioning

confidence: 99%

“…Two comprehensive reviews of a large number of existing methods have been presented in [1,2]. Vehicle license plate recognition is still a challenging issue for Iranian license plates [3][4][5][6][7].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Enhancing the license plates character recognition methods by means of SVM

Ghahnavieh

Amirkhani

Raie

2014

2014 22nd Iranian Conference on Electrical Engineering (ICEE)

View full text Add to dashboard Cite

Any vehicle license plate recognition system consists of three main components which include plate detection, character segmentation, and character recognition. The main theme of this paper is the improvement and innovation in the recognition of license plate characters. The assessments and the comparison between the performances of the methods proposed in this paper and the previous methods have all been done on one database and in similar software and hardware environments. The database includes about 20000 characters which have been extracted by license plate recognition systems from images obtained in real situations, i.e. day, night, and different distances and angles. This database contains low-resolution (20 x 12) characters, and also includes images with noise, distortion and omission. In a review and evaluation of different methods that use the support vector machine (SVM) to enhance the separation process, the MLP-SVM approach has been found to be one of the best. This method has recognized all the characters of a license plate with an accuracy of 95.86% in an average time of 92.9 ms. In this paper, to improve the character recognition, two methods based on combining probabilistic classifiers with the SVM have been proposed, and the outcome is the achievement of 96.7% accuracy and an average time of 57.4 ms.

show abstract

Section: B the One-against-one Methodsmentioning

confidence: 98%

mentioning

confidence: 99%