Abstract
This thesis has proposed and demonstrated a new 3D map which is realistic image-based, thus enabling geometry measurements and geo-location services. First and most important work of this study is the implementation of positioning and spatial measurements with image-based 3D maps, while current image-based 3D maps like Google Street View only provide virtual experience in terms of photos in which a number of details representing the topographic and terrain attributes are all lost, such as shapes and heights. Another contribution relates to quality analysis for a MRI 3D map. The research has deeply analyzed outlier impacts and introduced Dilution of Precision (DOP) values which are used to evaluate geometry quality for geo-referenced images either or measured vision points. In addition, the thesis has conducted a loose integration of mapping sensors like camera and GPS, and investigated image matching and stitching. The thesis has also proposed image-based 3D maps cooperated with Street View, in that panoramic viewing could make 3D maps more interactive with users, also bring an interesting immersive circumstance. This study has mainly focused on vision point positioning and implementation of spatial measurements on the MRI 3D map, analysis of data quality, outlier detection, and image processing. Users could obtain their orientation, position and object distance over basic functions of the MRI 3D map. A prototype geo-referencing system has been designed for the implementation of the MRI 3D mapping.