The Road Detection Algorithm In The Mobile Robot Visual Navigation

With the development of image processing technology and the enhancement of the ability of computers,vision-based mobile robot navigation system becomes an important topic recently in the realm of robot.Of all the key technologies,vision-based navigation is aiming to perceive and understand the surrounding so as to find out a safe path,to ensure the mobile robot to drive autonomously,steadily and safely without manual operation.The road detection system presented here is one of the key technologies in vision-based navigation. The main research contents are as follows:(1) This paper analyzes the advantages and disadvantages of 2 color space,which is RGB color space and HSI color space.It also compares several image pre-processing algorithms.Finally,we choose to enhance the image of the road through histogram equalization in HSI color space in the first place,and then smooth the image in RGB color space.These two steps formed the image pre-processing method.(2) The accurate of road region segmentation affects the following parts of the detection directly.In this paper,a wide range of threshold segmentation algorithms are studied in RGB and HSI color space.Through the experimental,comparison,a modified threshold segmentation method based on image entropy is selected,then combine with the mathematical morphology method to reduce the noise of the image.(3) This paper analyzes the anti-interference ability of the Freeman chain code,then a chain code tracking algorithm based on the region's area is presented to track the road edge.(4) DirectShow is applied to build a platform on which the real-time video processing and rendering are performed.A real-time video sequence capturing and data processing system based on DirectShow is designed and implemented in Visual C++6.0 environment.The emulation of all the algorithms are tested on this platform.Finally,experiments under different conditions are performed to verify the validity of the road detect system.The experimental results manifest that for the complex and varied road images,a better solution is provided in this paper,which guarantees the need of real-time,robust and correct for navigation and satisfies the practical requirement.