Vehicle information extracted from high resolution remote sensing image is of great significance in civil and military fields. To improve the accuracy and efficiency of the vehicle information extraction, the combined SURF (speeded up robust features) and support vector machine(SVM) algorithm is proposed to extract the vehicle information of the interest region. The edge information redundancy eliminating method and semi-search strategy are used to enhance the identification accuracy and reduce the amount of calculation. Vehicle information in the 0.25 m resolution remote sensing image of Nanshan District in Shenzhen is extracted and tested, the results show that the false rate is less than 20% and the extraction time can be controlled in minute level. The applicability of the method is demonstrated.