Pre-process of image of hazard recognition method based on single camera
- Paper number
IAC-11,A3,2.P,16,x10700
- Author
Mr. Yoshifusa Demizu, University of Tokyo, Japan
- Coauthor
Mr. Jianjun Zhu, Department of Engineering ,The University of Tokyo, Japan
- Coauthor
Prof. Tatsuaki Hashimoto, Japan Aerospace Exploration Agency (JAXA), Japan
- Year
2011
- Abstract
Landing mission plays significant role in space exploration for mining more information from moon. During landing procedure, automatic hazard detection and avoidance function should be improved to meet the need of future moon exploration missions. To cover the disadvantages (such as heavy, time costing and high energy consuming) of dimension elevation map which is constructed from laser range data, we have proposed a new method of hazard recognition by utilizing shadows, terrain features, and some priori information such as sun angle and geological knowledge of an image, instead of constructing a three dimensional elevation map. The differences of brightness in an image offer part of the information of craters and rocks. Based on the shadow and angle of sunlight hazards could be recognized. However, the simulation based on the image from LRO shows the modest accuracy performance of the method. Thus, a suitable way of image pre-process is required to improve this method. We found the craters and rocks which exposed to the sun with regard to the image from LRO have obvious different brightness with the ones of the areas around the craters or rocks. Also, the gray intensities of two sides of the obstacles were mostly different between each other. For instance, the side of rocks that orient to the sun is has the highest gray intensities. Opposite to the side orient to the sun, the one back to the light has the lowest intensities. The most important thing is that the gray intensities of the area around these rocks almost equal to the average gray intensities of whole image. Compare to the average gray intensity, only the area which has relative high gray intensities and relative low gray intensities beside the same area simultaneously could be considered as an obstacle and do the next step to extract the edge of the obstacle. To avoid misrecognition, the priority value of high gray intensity threshold which could be calculated will be higher than average gray intensity and also the low gray intensity threshold will be lower than average gray intensity. The simulation was based on the image from artificial planetary environment and LRO. Proposed method correctly detected and divided the craters and rocks in the image. From the simulation result, the pre-process of the image from LRO shows a good accuracy performance of hazard detection of method based on shadows and angle of sunlight.
- Abstract document
- Manuscript document
(absent)