Terrain classification in field environment with grid-based SURF feature

Mobile robot´s terrain classification in field environment is severely affected and obstructed by the changes of the external conditions. This paper introduces a kind of feature which is based on SURF (Speed-Up Robust Features), then we improve and optimize it to make it suitable for terrain classification and we call it grid-based SURF feature; besides we apply the grid-based SURF feature in terrain classification in field environment. Compared with other traditional features such as color histogram and color moment, the experiments result shows that grid-based SURF feature can not only get a good classification result but also keep being better than others when illumination changes, thus the grid-based SURF feature is applicative and robust in terrain classification in field environment.