Vehicle detection from video sequence based on gabor filter

Detecting vehicles from video sequence is very challenging due to the wide varieties of vehicle appearances and the complexity of the backgrounds. At present, many algorithms in the image recognition have a narrow applicability and a weak real-time. Aiming at this problem, a recognition method which was combined by features extraction using Gabor wavelet and BP neural network algorithm for the classification of vehicle types based on the texture model is proposed. Firstly, the background image is captured and automatically updated. Moving vehicles is detected and abstracted by background subtraction. For removing the noise, the images were filtered by median filtering. Secondly, a model of feature vector based on Gabor filter is built. At last, BP neural network classifier was designed to train and identify the feature points. Experiments show that this approach has better robustness, good real-time and high recognition rate, with a wide range of practical applications.