Three-Dimensional Coordinate Reconstruction of Each Digital Image Pixel in a Simple Stereo Configuration Using the Belief Propagation Method

EasyChair Preprint 5741

Abstract

One of the major problems in computer vision is how to reconstruct the threedimensional  coordinates of each pixel which are shown in a digital image (inverse problem: each pixel to restore its natural coordinate). If we can determine the disparity of a pixel, we can determine the position of the pixels on the threedimensional coordinate system using only simple calculations. The research was able to approach the disparity of almost every pixel with the actual disparity. This could be be seen by comparing the digital image disparity results with actual digital image.

Keyphrases: Disparity, Segmentation, correspondency

@booklet{EasyChair:5741,
  author    = {Efron Manik},
  title     = {Three-Dimensional Coordinate Reconstruction of Each Digital Image Pixel in a Simple Stereo Configuration Using the Belief Propagation Method},
  howpublished = {EasyChair Preprint 5741},
  year      = {EasyChair, 2021}}