VPL: A View Planning Library for Automated 3D Reconstruction

During the last decade, automated three-dimensional (3D) object reconstruction and inspection have become widely used in the same way that 3D sensors and positioning systems have become affordable. For example, nowadays drones or underwater vehicles are capable of building 3D models of complex structures. To achieve an autonomous reconstruction it is essential to determine each pose of the sensor, from where an observation is made. In the literature, this task is called view planning. To develop a custom application or propose a new view planning algorithm requires several software tools that are usually reimplemented each time; this causes an effort duplication and is one of the reasons why view planning research papers usually do not present comparisons. We present VPL, a software framework to develop view planning algorithms and compare their performance versus other approaches. The library is written in C++ and it is released under open source BSD license.