Modeling Instantaneous Changes In Natural Scenes - Computer Science > Computer Vision and Pattern RecognitionReport as inadecuate

Modeling Instantaneous Changes In Natural Scenes - Computer Science > Computer Vision and Pattern Recognition - Download this document for free, or read online. Document in PDF available to download.

Abstract: This project aims to create 3d model of the natural world and model changesin it instantaneously. A framework for modeling instantaneous changes naturalscenes in real time using Lagrangian Particle Framework and a fluid-particlegrid approach is presented. This project is presented in the form of aproof-based system where we show that the design is very much possible butcurrently we only have selective scripts that accomplish the given job, acomplete software however is still under work. This research can be dividedinto 3 distinct sections: the first one discusses a multi-camera rig that canmeasure ego-motion accurately up to 88%, how this device becomes the backboneof our framework, and some improvements devised to optimize a know frameworkfor depth maps and 3d structure estimation from a single still image calledmake3d. The second part discusses the fluid-particle framework to model naturalscenes, presents some algorithms that we are using to accomplish this task andwe show how an application of our framework can extend make3d to model naturalscenes in real time. This part of the research constructs a bridge betweencomputer vision and computer graphics so that now ideas, answers and intuitionsthat arose in the domain of computer graphics can now be applied to computervision and natural modeling. The final part of this research improves upon whatmight become the first general purpose vision system using deep beliefarchitectures and provides another framework to improve the lower bound ontraining images for boosting by using a variation of Restricted Boltzmannmachines RBM. We also discuss other applications that might arise from ourwork in these areas.

Author: Vikram Dhillon



Related documents