3D Object Reconstruction From 2D Point Clouds Using Multi-stage Edge Detector

The research, as described in this paper, is focused on finding a low-cost alternative to modern 3D reconstruction technologies used to recreate objects from orthogonal images. The main objective of the research is to create a 3D shape of an object, applying various software filters (C++ software environment) on six principal views of the object (bottom, top, front, back, left, and right sides), captured as RGB images and utilizing a mathematical algorithm (Python software environment) for final processing. The outputs of the filters are 2D Point Cloud estimations, which are then unionized into the single 3D Point Cloud estimation with the help of calculations in Python code. Generated Point Cloud estimations are then compared to existing 3D CAD models to performa visual inspection to check the performance ofthe algorithm. Additionally, mathematical formulas are used to get the error value in order to check how similarly received estimations are comparing to the actual model. The estimator is tested with total of twenty sample objects.