| China has a vast territory.It is of great significance to quickly obtain 3D geographic information and timely carry out long-distance off-road path planning for military reconnaissance,earthquake disaster emergency rescue and regional economic construction.The traditional 2D plane map cannot reflect the topographic information of the areas and not effective information support for study area.Digital Elevation Model(DEM)can reflect the features of the terrain,but single DEM coverage is limited.Since multi-stripe DEM system error exists,it is difficult to directly set.As a result,long-distance path planning based on DEM has become a difficulty in DEM application at present.Long-distance path planning faced the problem of low efficiency of time in 3D environment.It has the adverse impact on the completion of off-road tasks.Therefore,to carry out the access to the areas geography 3D information,and fast long-distance off-road path planning research has very important significance.Space-borne Synthetic Aperture Radar(SAR)has the advantages of fast performance and regardless of weather.Interferometry synthetic aperture radar(In SAR)technology is widely used to generate DEM to obtain geographic information.Due to the influence of SAR side view imaging characteristics and satellite orbit,the coverage of single DEM generated based on In SAR is limited.In order to obtain a large-scale of complete geographical 3D information in the area,and to solve the problem that multi-stripe DEM mosaicking is difficult in the area in terms of topographic error,this paper proposes a fast mosaicking method of multi-stripe DEM generated by In SAR,and improves the efficiency of mosaicking by using parallel computing.The areas of a large-range DEM,which is basis of the 3D geographic information,is obtained to carry out long-distance off-road tasks.The priority is to make the path planning in the shortest possible time.The efficiency of traditional path planning is low in long-distance off-road path planning tasks.And map scale is larger,the more complex topography,the lower the planning efficiency.Aiming at the problem of long processing time in long-distance off-road path planning,this paper proposes a method of long-distance off-road path planning based on terrain data map,which greatly improves the efficiency of long-distance off-road path planning.This paper conducts research on multi-strip fast mosaicking and fast long-distance off-road path planning of DEM generated by In SAR.The main research is as follows:1)Aiming at the problems that the multi-stripe DEM generated by In SAR is difficult to directly mosaic due to the influence of topography,error and other aspects,and the execution efficiency of multi-stripe DEM mosaicking is low.In this paper,a fast and large-scale DEM multi-stripe mosaicking method is proposed.Firstly,according to the causes of multi-stripe DEM errors,the errors are divided into plane errors and elevation errors.Then,the elevation control points and plane tie-points are extracted respectively.In the elevation direction,the uniform elevation control points are automatically extracted from the grid based on the selected laser elevation control points.In the plane direction,the sliding matching method is used to extract the plane tie-points in the DEM overlapping region.The proposed self-classification method is used to match the plane tie-points in the multi-image overlapping region.Secondly,the elevation control points and the plane tie-points are used for block adjustment in the direction of elevation and plane respectively.After the systematic errors in the direction of the elevation and plane are corrected,a whole large-scale DEM was generated by the mosaicking strategy of first along-track followed by cross-track based on the feathering mosaicking method.In addition,to solve the problem of time-consuming mosaicking,the CPU/GPU hybrid acceleration strategy is adopted to improve the execution efficiency.Experimental results show that the efficiency of the proposed CPU/GPU hybrid acceleration strategy is nearly 20 times higher than the original processing efficiency,and the multi-strip fast mosaicking of DEM generated by In SAR is realized.2)Aiming at the low efficiency of long-distance off-road path planning processing,this paper proposes an improved A-Star algorithm.Based on the research of fast mosaicking method of In SAR-generated multi-stripe DEM,a complete large-scale terrain data map is constructed based on 3D geographic information obtained from the areas.The improved A-Star algorithm is developed for long-distance off-road path planning tasks without road network to determine the feasible path between the starting point and the end point.In this study,the algorithm is optimized from two aspects of data structure and retrieval strategy.First,the hybrid data structure of minimal heap and two-dimensional array greatly reduces the time complexity of the algorithm.Secondly,an optimized search strategy is designed,which improves the execution efficiency by not checking whether the target is reached at the initial stage of searching the optimal path.To evaluate the efficiency of the proposed algorithm,three different off-road path planning tasks were designed,namely short-,medium-and long-distance.Each task was corresponding to three different off-road vehicles.A total of nine groups of experiments were carried out,and the results show that the processing efficiency of the proposed algorithm is significantly better than the traditional A-Star algorithm.Compared with the traditional A-Star algorithm,the path planning efficiency of the improved A-Star algorithm is at least 4.6 times higher,the maximum acceleration reaches 550 times in the long-distance off-road path planning task.Experimental results show that the improved algorithm greatly improves the efficiency of long-distance off-road path planning. |
PDF Full Text Request