Research On Data Registration And Integration Of Terrestrial Laser Scanning Point Clouds And Fisheye Panoramic Images

Multi-sensor information integration and multi-source heterogeneous dataintegration is an important development direction of current remote sensingtechnology. Integration of terrestrial laser scanning point clouds and fisheyepanoramic images has very important significance for the perception andunderstanding of the large-sight environment, the virtual scene analysis. This paperpresents a data-oriented method of the data integration of laser point clouds andfisheye panoramic images: by the cylinder model perspective image transformation isused to achieve the same reference system; through the line features extraction andregistration of two data the three-dimensional alignment under the constraint-basedviewpoint is realized; through the establishment of a virtual point, two data forms ofmixed color discrete points and panoramic images with geo-referenced are used toexpress the scene information, and finally by the metadata expansion of viewpointclass and space model class data-level integration of both is achieved. Papers haveconducted the integration experiments of terrestrial laser point clouds and fisheyeimages. Experimental results show that proposed method has the ability to integratemultiple sources data, and provides the data condition for the integration, databaseand cross-platform of the large-scale and multi-scene data.