Utilization of Photosynth Point Clouds for 3D Object Reconstruction


There is a growing demand in modeling amorphous shapes like statues, figurine or monuments for computer visualization and documentation. Using photographs as the data source is a most convenient recording mode on site. The equipment is easy to handle and transportation is in no case a problem as it can occur if a laser scanning device has to be employed.

A photograph is a container of high information density. It carries radiometric information and can provide range values as well. In terms of computer vision, structure from motion is a process to find the correspondence between images. Features must be tracked from one image to the next. The 3D positions of the feature points and the camera movement is the result of the registration process. In the cultural heritage community the Epoch ARC 3D Web service is in common use. Linked with the open source tool MeshLab it provides an automated workflow including object reconstruction, mesh processing and textured rendering.

Recent developments from Microsoft introduce photo browsing. The Web community can view images from cities in Virtual Earth and participate with own objects applying Photosynth. Photosynth is designed as an image browser for objects, documented by internet imagery. The user navigates through a bundle of images representing the object. A smooth transition from one photo to the next is leading to the impression of a 3D model. Photo positions are known as well as a point cloud emerged from the registration process.

Applying a network protocol analyzer provides the location of the binary point cloud files. A Python script affords conversion into common CAD formats like PLY or DXF. Producing a low level polygon mesh for real time visualization exacts selecting the points of interest and mesh processing. The latter task will be solved applying MeshLab. It has to be mentioned, that the focus in this context points more on that low level polygon model instead of a high accurate and high density model. Furthermore Blender with its comfortable function for UV texture mapping and export script to X3D will be applicable to construct a complete 3D scene. This contribution introduces a 3D object reconstruction method from a series of photographs processed with Web tools and open source software.

If you are you interested in this seminar topic, mail to us to get

the full report * of the seminar topic.

Mail ID: - contact4seminars@gmail.com 

* conditions apply

– OR –

Click here for Quick Contact (Request for Topics)

  • © 2008 – 2013 seminars4you,

  • All rights reserved.