Research On Photogrammetric Processing For Mars Topographic Mapping

Topographic mapping is the basis for scientific research of Mars.High precision topographic information is essential in selecting landing sites and assisting safe landing.Due to the far distance between Mars and Earth,we can't use GPS to determine satellite orbits.Besides,no control points can be used to geo-position Martian surface stereo imagery.Meanwhile,the satellite imagery of Martian surface has little texture information.These factors make the topograpihc mapping of Mars become a challenging problem.In this thesis,based on the analysis of the key technologies of Mars topographic mapping,the geometric processing of Mars satellite imagery,combined adjustment of satellite imagery and laser altimetry data,dense image matching and automatic extraction of digital elevation model are investigated.The main contents and innovations of the paper are summarized as follows:1.The background and current research of Mars topographic mapping are discussed and typical exploration tasks related to Mars topographic mapping are analyzed.After introducing the current research of China's lunar topographic mapping,the main problems for China's Mars topographic mapping are pointed out.2.The coordinate system,map projection and height datum for Mars topographic mapping are discussed.The planetary data processing system utilized by USGS are analyzed.The photogrammetric data preprocessing of Mars Express HRSC imagery and Mars Reconnaissance Orbiter Hi RISE imagery are researched and the rigorous geometric model of HRSC imagery is established.3.The coordinate system definitions and transformation among different coordinate systems of satellite imagery are researched.A fast algorithm for back projection of ground points based on best scan line searching is proposed.The fast algorithm for back projection can be used in geometric correction,image fusion and image matching.Experiment results demonstrate that based on the proposed algorithm,the geometric correction efficiency of satellite imagery can be improved greatly.4.The bundle adjustment of satellite imagery and automatic generation of tie points are researched.To take both advantages of high accuracy laser altimetry data and high resolution satellite imagery,laser altimetry data are incorporated into the bundle adjustment as additional measurement which is helpful to improve the geopositioning accuracy of Martian surface imagery.Based on orientation fix model and quaternion theory,the combined adjustment mathematical model is established and the corresponding error equations are derived.5.The geometric distortion and scale change have great influence on image matching,thus it is not appropriate to perform image matching on original imagery which has great geometric distortion.An approximate epipolar image resampling method for HRSC imagery based on horizontal imagery is put forward and the epipolar geometry of HRSC is analyzed.6.A new image matching method for Martian surface satellite imagery based on Precise Point Prediction Model(P3M)is proposed.A detailed elaboration is given on the basic concepts of densification matching by levels and how to select good feature points and use affine transformation to construct precise point prediction model.The point prediction accuracy and robustness of our matching method are analyzed and experiment results demonstrate the method's feasibility.7.Based on precise point prediction matching concept,a method for automatic extraction of digital elevation model is designed.The extracted DEM is compared to ESA's products and the systematic errors between them are analyzed.8.For acquiring high precision digital elevation model on landing sites,a pixel-wise image matching method is invented and the parallel computing strategy on multi-core platform is designed.Experimental results demonstrate the method's feasibility.