Real Time 3D Reconstruction from Monocular Video

November

1st week

  • Preparing some posters and other graphic materials to expose in AR&PA 2014 (to be held in two weeks)
  • Studying the interesting solution for colume cost computation using octrees from @Steinbr2014 titled "Volumetric 3D Mapping in Real-Time on a CPU"

2nd week

  • My third (and last) class of OpenCV with Python. Unfortunatley, there is no time to explain segmentation examples which are not necessary to complete the practice.
  • Writing the formal statement of the OpenCV practices for the students
  • Assisting to AR&PA 2014. I meet very interesting works in the fields of mobile mapping (based on LIDAR devices and spherical panoramas) and real time rendering of previously reconstructed scenes using multiples views. The reference product in the segment of photogrammetry is Agisoft PhotoScan

3rd week

  • Augmenting the framework to include scene updating and maintenance in separate threads. The same idea has been used previously in the PTAM implementation
  • Compiling and testing the Open Source alternative to PhotoScan, called MicMac, a project from the IGN Francaise. It is a good alternative but it is not good as PhotoScan, specially in regard to the image calibration step.

4rd week

  • Collaborating with my lab partners in writing and reviewing the extended abstract for 3D ARCH 2015 titled: "AUGMENTED REALITY TO PRESERVE HIDDEN VESTIGES IN HISTORICAL CITIES. A CASE STUDY"
  • Installing and setting a ShareLatex environment to allow collaborative editing in our own server following the ISPRS guidelines for authors.
  • Reviewing slides related with my PhD work for the next meeting of the Asturian Cluster of Digital Contents in Gijon (thursday and friday)

References