A Stereo Visual Obstacle Detection Approach Using Fuzzy Logic and Neural Network in Agriculture

摘要

This paper describes a stereo visual obstacle detection approach using fuzzy logic and neural network in agriculture. Firstly, by training obstacle and non-obstacle colour features with a 6-input neural network, appearance_based method is applied to extract novel regions in monocular image. We apply improved FAST algorithm to extract features, which adapts illumination variation better. At last, we match feature points in the image pairs and obtain the disparity information. The results show that the algorithm is capable of detecting obstacles such as vehicles, humans, trees, plastic bar and other common obstacles in agricultural environment successfully.