Two-stage License Plate Detection using Gentle Adaboost and SIFT-SVM Wing Teng Ho, Hao Wooi Lim, and Yong Haur Tay Computer Vision & Intelligent Systems (CVIS) group, Universiti Tunku Abdul Rahman (UTAR). 9, Jalan Bersatu 13/4, 46200 Petaling Jaya, Selangor, Malaysia. email: wingtengh@gmail.com, zybler+rnd@gmail.com, tayyh@utar.edu.my Abstract This paper presents a two-stage method to detect license plates in real world images. To do license plate detection (LPD), an initial set of possible license plate character regions are first obtained by the first stage classifier and then passed to the second stage classifier to reject non-character regions. 36 Adaboost classifiers (each trained with one alpha-numerical character, i.e. A..Z, 0..9) serve as the first stage classifier. In the second stage, a support vector machine (SVM) trained on scale-invariant feature transform (SIFT) descriptors obtained from training sub-windows were employed. A recall rate of 0.920792 and precision rate of 0.90185 was obtained. Keywords: License plate detection, SIFT, SVM, Adaboost. 1. Introduction Vehicle license plate detection (LPD) is an important step in any automated vehicle verification system. It is used to search for license plate portion on an image before verification of the license plate can be done. Failure is imminent if the license plate region is not detected or wrongly found. There are many applications for vehicle verification system, such as stolen vehicles detection [10], driver navigation support [10], automated parking attendant [6], border crossing control [6], petrol station forecourt surveillance [6], personalized service via customer identification [6], automated toll ticketing [14] and etc. There are many constrains and difficulties in LPD when it comes to dealing with unconstrained environment such as scale, rotation, affine transformation, illumination, occlusion, translation; shearing, distortion and skew. In this paper, an approach that combined Adaboost and Support Vector Machine (SVM) in LPD is presented. Our framework will use Adaboost to do the character detection, and SVM is used to filter the false positive. 2. Background Adaboost has been a successful approach for face detection that minimizes the computational time and obtaining high detection rate. Viola et al. [15] presented a framework for face detection that achieves high detection rate rapidly. Motivated by [5], they introduced an image representation known as the integral image that allows the features used in the detection to be calculated in many scales or locations in constant time. They applied Haar-like features [15] to classify the patterns for an image, which are reminiscent of Haar wavelet used by Papageorgiou and Poggio [4][5]. Then, they used AdaBoost learning algorithm to select a simple Haar-like features from the over-complete features set. Chen and Yuille [20] demonstrated an algorithm for detecting text in natural images, also based on AdaBoost. They claimed that the set of features used for face detection by Viola and Jones [15] might not be suitable for detecting text. This is because there is less spatial similarity for the text compare to face; a face can be regarded as spatial similar object since it consists of facial features such as eyes, nose, and mouth that are approximately the same spatial position for any face. Some of the algorithms are designed specifically for detecting character such as the adaptive algorithm for text detection from natural scenes by Gao and Yang [8]. They developed a prototype system that can recognize Chinese sign inputs. Shapiro et al. [18] have introduced image-based vehicle license plate recognition (CLPR) system. The system consists of few processes such as the license plate localization that is used to locate the license plate for an image. In [7], they proposed a LPD method that implements a global 2009 First Asian Conference on Intelligent Information and Database Systems 978-0-7695-3580-7/09 $25.00 © 2009 IEEE DOI 10.1109/ACIIDS.2009.25 109 2009 First Asian Conference on Intelligent Information and Database Systems 978-0-7695-3580-7/09 $25.00 © 2009 IEEE DOI 10.1109/ACIIDS.2009.25 109