The Object Projection Feature Estimation Problem in Unsupervised Markerless 3D Motion Tracking

TL;DR

This paper introduces a novel real-time 3D motion tracking system that requires no prior knowledge of the object and works with a single low-cost camera, handling various object complexities and occlusions.

Contribution

It presents a new system for 3D motion tracking that estimates object features without object models or specialized hardware, enabling widespread application.

Findings

Operates in real time on standard cameras

Handles non-rigid, occluded, and motion-blurred objects

Does not require prior object knowledge or specialized hardware

Abstract

3D motion tracking is a critical task in many computer vision applications. Existing 3D motion tracking techniques require either a great amount of knowledge on the target object or specific hardware. These requirements discourage the wide spread of commercial applications based on 3D motion tracking. 3D motion tracking systems that require no knowledge on the target object and run on a single low-budget camera require estimations of the object projection features (namely, area and position). In this paper, we define the object projection feature estimation problem and we present a novel 3D motion tracking system that needs no knowledge on the target object and that only requires a single low-budget camera, as installed in most computers and smartphones. Our system estimates, in real time, the three-dimensional position of a non-modeled unmarked object that may be non-rigid, non-convex, partially occluded, self occluded, or motion blurred, given that it is opaque, evenly colored, and enough contrasting with the background in each frame. Our system is also able to determine the most relevant object to track in the screen. Our 3D motion tracking system does not impose hard constraints, therefore it allows a market-wide implementation of applications that use 3D motion tracking.