Research Progress of Obstacle Detection Based on Monocular Vision

In recent years, unmanned autonomous vehicle navigation technology or UAV research is a hot topic in the field of machine vision, and obstacle detection is the basis of autonomous navigation. In this thesis, the author analyzed the deficiency of stereoscopic vision and classified current monocular obstacle detection algorithms. According to the properties of algorithms, this paper elaborated the function of algorithms based on features, focal length and movements respectively. The author also discussed current situation and prospects for various types of algorithms, including feature extraction and matching, focusing and defocusing model, background subtraction, frame difference and optical flow. Finally, this paper summarized advantages and disadvantages of different autonomous navigation algorithm and discussed the prospect of the monocular obstacle detection technology.