This paper is mainly about the recognition of car license plate characters. A method based on two-kind feature and two-stage classifier is presented. For car license plate character recognition, there are two kinds of features that can be extracted: configurable feature and statistical feature. Usually, the classifier whose inputs are statistical features is easy to train, but its robustness isn't good. The advantage of the classifier whose input is configurable feature is its better reliability, but this kind of classifier usually needs a complicated pretreatment process. So, the classifier, which based on two-kind feature and two-stage classifier, synthesizes the advantages of the two kinds of classifiers and avoids the flaws. The two classifiers in this paper are both trained by SVM. Also, the experiment results show that the recognition rate is higher, and that multi-stage classifier is obviously superior to single classifier.
[1]
Zi-Quan Hong,et al.
Algebraic feature extraction of image for recognition
,
1991,
Pattern Recognit..
[2]
Qiao Jin.
Multi stage Classifier for Unconstrained Handwritten Numerals Recognition
,
1999
.
[3]
Gu Yang.
Digit Character Recognition Based on Wiener Filter, Karhunen-Loeve Transform and BP Network
,
2002
.
[4]
Sheng Ji.
A Hybrid Algorithm of Vision Learning and Neural Networks for Recognition of Handwritten Digits
,
2001
.
[5]
Hu Zhong,et al.
HANDWRITTEN DIGIT RECOGNITION BASED ON MULTI-CLASSIFIER COMBINATION
,
1999
.
[6]
He Xing.
Application of BP neural network algorithm in number recognition
,
2000
.