Study On The Recognition Method Of Lunar Stone Based On Machine Vision

Chinese Government published Chinese Spaceflight White Book in November 2000. The research on deep space exploration especially the moon exploration project was established. This filled up the blank in the deep space exploration for China and founded new science basis for human beings to set up the moon bases. China has carried out its first plan of lunar exploration Chang E Project since March 1st, 2003. This project is carried out through three steps. The last two steps of the project will be carried out in 2012, when the lunar rover will play its role.Study on the recognition methods of lunar stone based on machine vision is just started in China, so the technologies are not very mature. Along with the development of the moon exploration of China, the lunar rover will become a hot spot of science research. There are two parts in the moon exploration of China: the autonomous navigation and the ground control. It is the key technology to recognize the obstacles for the lunar rover to navigate autonomously. So a set of steady vision system is an essential part of lunar rover.Image segmentation is the most difficult and important pre-processing technology in the field of pattern recognition and computer vision. It has been widely researched and used. This technique both involved in the data processing and the knowledge expression. So it is the classical difficult problem of image information engineering. And it has been the bottle-neck problem and technology difficulty of pattern recognition for a long time.The moon surface environment is complicated and variable. Especially the refraction makes the environment even more complex. It brings great difficulty to image segmentation. Meanwhile because of the limit of image segmentationalgorithms, there is no universal segmentation algorithm to fit all the environments. So it needs to find image segmentation algorithms considering multiple factors. Then one algorithm fitting one environment can be realized. Thus the lunar rover can choose according algorithms to recognize the right paths, and autonomous navigation can be realized.The title of this paper is Study on the Recognition Methods of Lunar Stone Based on Machine Vision. The paper mainly consists of four parts: 1.The segmentation algorithms of stone images in different environments. 2. The comparison and analysis of all the segmentation algorithms. 3. The location and recognition of stones. 4. The experiments and conclusions.Multiple dynamic threshold segmentation algorithms aimed at different illuminations and different environments are studied, and the principles are studied and explained particularly. There are the algorithms based on Mathematical Morphology grads operator; image segmentation algorithms based on Fisher Criterion Function; image segmentation algorithms based on 2D maximum between cluster variance; image segmentation algorithms based on the maximum correlation criterion; image segmentation algorithms based on texture information; image segmentation algorithms based on partial threshold integrating whole threshold and so on.Lots of experiments images are illustrated. The segmentation algorithms are compared and analyzed. Thus the characteristics of the algorithms are obtained. Thus which algorithm can be used in a certain environment can be decided.Morphology processing algorithms are proposed to deal with binary images aiming at the features of the stones. It mainly contains: the elimination of isolated point; Mathematical Morphology dilation and erosion; the area marking algorithms of binary images. Thus a good foundation of the following location and recognition can be established. The location of the stones is realized by projection method.All the algorithms are integrated into a system-software on Visual C++ platform.Actual experiments are carried out in the lunar rover proving ground of the ChinaAcademy of Space Technology. Experimental results show that the algorithms are effective in different illuminations and different environments.