Abstract
Abstract. This paper explores a method of generating panoramic street strip image map which is called as “Pano-Street” here and contains both sides, ground surface and overhead part of a street with a sequence of 360° panoramic images captured with Point Grey’s Ladybug3 mounted on the top of Mitsubishi MMS-X 220 at 2m intervals along the streets in urban environment. On-board GPS/IMU, speedometer and post sequence image analysis technology such as bundle adjustment provided much more accuracy level position and attitude data for these panoramic images, and laser data. The principle for generating panoramic street strip image map is similar to that of the traditional aero ortho-images. A special 3D DEM(3D-Mesh called here) was firstly generated with laser data, the depth map generated from dense image matching with the sequence of 360° panoramic images, or the existing GIS spatial data along the MMS trajectory, then all 360° panoramic images were projected and stitched on the 3D-Mesh with the position and attitude data. This makes it possible to make large scale panoramic street strip image maps for most types of cities, and provides another kind of street view way to view the 360° scene along the street by avoiding the switch of image bubbles like Google Street View and Bing Maps Streetside.
Highlights
This paper explores a method of making panoramic street strip image map which is called as “Pano-Street” here and contains both sides, ground surface and overhead part all directions of a street with a sequence of 360° panoramic images which are captured with a Point Grey’s Ladybug3 mounted on the top of Mitsubishi Mobile Mapping System (MMS)-X 220 at 2m intervals along the streets in urban environment (Chen et al, 2012), and projected/stitched on to one special 3D mesh called projection base plan (See Figure 1) which is equivalent to the DEM/DSM used to generate ortho-image with aerial/satellite images, and constructed with existing 2D/3D GIS data along the street
The purpose of this paper is to provide a kind of low-cost, shortcut and useful GIS data making method to meet the requirements of navigation, driving direction pre-visualizations and augmented reality as demonstrated in Google Earth or Microsoft Virtual Earth with our Mobile Mapping System (MMS)
In the last decades, much more efforts have been invested into large scale urban 3D model construction from vision-based 3D modelling and fusion 3D laser point clouds with images for these applications such as navigation, driving direction previsualizations and augmented reality as demonstrated in Google Earth or Microsoft Virtual Earth, but the progress on the street view level is not so amazing notable and the achieved results are still on the level of research
Summary
Much more efforts have been invested into large scale urban 3D model construction from vision-based 3D modelling (Baillard et al, 1999; El-Hakim, 2002; Bentrah et al, 2004a; Surendra et al, 2014) and fusion 3D laser point clouds with images (Frueh et al, 2005; Zhao and Shibasaki, 2001; Stamos and Allen, 2000) for these applications such as navigation, driving direction pre-visualizations and augmented reality as demonstrated in Google Earth or Microsoft Virtual Earth in the last decades, but the progress on the street view level is not so remarkable and the achieved results are still on the level of research. Two kinds of low-cost, shortcut and functionary methods to provide users much more detail immersive street scenes are systems such as Google Street View and Microsoft Live StreetSide that enable users to virtually visit cities by navigating between immersive 360° panoramic images (Vincent 2007), or bubbles, and multi-perspective strip panoramas that can provide a visual summary of a city street with a long image strip constructed a sequence video images along the street (Román et al 2004; Román and Lensch 2006; Agarwala et al 2006; Rav-Acha et al 2008) The former could provide a photorealistic impression from a particular viewpoint inside a bubble by panning and zooming the bubble or the image, but they do not provide a good visual sense of a larger aggregate such as a whole city block or longer street and need vast amount of geo-registered panoramic imagery. Because all of the source imagery is compressed into a single flat summary, they lack the spatial geographic information as the traditional ortho-images generated from aero and satellite images
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
