License plate detection based on improved Real Adaboost

License plate recognition (LPR) which has numerous applications in our lives has been widely adopted in intelligent transport system. In this paper, several improvements have been made to the Real Adaboost based real-time license plate detection algorithm, which is the first and important step in LPR. Firstly, apart from the normally used Haar-like feature, the local edge orientation histograms (EOH) feature is introduced, which provides an over-complete license plate feature set can decrease the training number and enhance the detection rate. Secondly, weak classifiers in improved Real Adaboost are configured as confidence-rated look-up-table (LUT) of Haar-like feature and EOH feature. The LUT type weak classifier is constructed through sigmoid function has advantages of smooth and flexibility. The experiments show that the proposed method is more efficient and accurate than the original Real Adaboost algorithm.