RGB-D Sensors Data Quality Assessment and Improvement for Advanced Applications

Abstract

Since the advent of the first Kinect as a motion controller device for the Microsoft XBOX platform (November 2010), several similar active and low-cost range sensing devices, capable of capturing a digital RGB image and the corresponding Depth map (RGB-D), have been introduced in the market. Although initially designed for the video gaming market with the scope of capturing an approximated 3D image of a human body in order to create gesture-based interfaces, RGB-D sensors’ low cost and their ability to gather streams of 3D data in real time with a frame rate of 15–30 fps, boosted their popularity for several other purposes, including 3D multimedia interaction, robot navigation, 3D body scanning for garment design and proximity sensors for automotive design. However, data quality is not the RGB-D sensors’ strong point, and additional considerations are needed for maximizing the amount of information that can be extracted by the raw data, together with proper criteria for data validation and verification. The present chapter provides an overview of RGB-D sensors technology and an analysis of how random and systematic 3D measurement errors affect the global 3D data quality in the various technological implementations. Typical applications are also reported, with the aim of providing readers with the basic knowledge and understanding of the potentialities and challenges of this technology.