• Document: SVM Based License Plate Recognition System
  • Size: 865.94 KB
  • Uploaded: 2019-02-14 00:41:19
  • Status: Successfully converted

Some snippets from your converted document:

2010 IEEE International Conference on Computational Intelligence and Computing Research SVM Based License Plate Recognition System Kumar Parasuraman, Member IEEE and Subin P.S cannot be recognized due to very poor illumination, motion Abstract— In this paper, we review the use of support vector blurred effect, fade characters and so forth. Furthermore, all machine concept in license plate recognition. Support vector the methods abovementioned performed license plate machines (SVMs) are a set of related supervised learning recognition after characters had been segmented. However, methods used for classification and regression. In simple words, images taken in real-time may be difficult for character given a set of training examples, each marked as belonging to one segmentation due to poor image quality. Improperly of two categories, an SVM training algorithm builds a model that predicts whether a new example falls into one category or the segmented characters will result in misrecognized characters. other. Intuitively, an SVM model is a representation of the In order to improve the recognition system performance, we examples as points in space, mapped so that the examples of the propose a new SVM-based multiclass classifier to recognize separate categories are divided by a clear gap that is as wide as number plates with poor quality. The number plates are possible. New examples are then mapped into that same space recognized without going through character segmentation. and predicted to belong to a category based on which side of the gap they fall on. Here we are using the concept of SVM in LPR II. SVM BASED MULTI-CLASS CLASSIFIER systems. Then a number plate recognition algorithm is proposed for character segmentation and recognition. This algorithm Since 1960s SVMs have become more and more important employs an SVM to recognize numbers. The algorithm starts in the field of pattern recognition. SVM [14, 15] is forcefully from a collection of samples of numbers from number plates. competing with many methods for classification. An SVM is a Each character is recognized by an SVM, which is trained by supervised learning technique. SVM takes Statistical Learning some known samples in advance. In order to recognize a number Theory (SLT) as its theoretical foundation, and the structural plate correctly, all numbers are tested one by one using the risk minimization as its optimal object to realize the best trained model. The recognition results are achieved by finding generalization. They are based on some simple ideas and the maximum value between the outputs of SVMs. Multi-class SVMs are developed to classify the given number plate provide a clear intuition of what learning from examples is all candidate. The experimental results show that our new method is about. More importantly, they possess the feature of high of higher recognition accuracy and higher processing speed than performance in practical applications. The SVMs use using traditional SVM based multi-class classifier. This new hyperplanes to separate the different classes. Many approach provides a good direction for automatic number plate hyperplanes are fitted to separate the classes, but there is only recognition. Here we can conclude SVM is better than any other one optimal separating hyperplane. The optimal one is supervised learning. expected to generalize well in comparison to the others. A new data sample is classified by the SVM according to the Keywords- Automatic Number Plate Recognition, Support decision boundary defined by the hyperplane. Among many vector machine. classification methods, SVM has demonstrated superior performance. It has been successfully utilized in handwritten I. INTRODUCTION numeral recognition [11, 12]. However, SVM was originally designed for binary classification, and its extension to solve A number pla

Recently converted files (publicly available):