Disparity Estimation Techniques
Mounir Kaaniche, Raffaele Gaetano, Marco Cagnazzo, Béatrice Pesquet‐Popescu · 2013
The disparity estimation problem, also known as the stereo matching problem, has been extensively studied in computer vision, and surveys of the different techniques proposed in the literature can be found in Scharstein and Szeliski and Brown. Traditionally, two approaches, block-based or pixel-based, can be used to estimate the disparity field. This chapter presents the geometrical models for stereoscopic imaging. It first introduces the basic concepts of camera models, and then establishes the geometrical relation between the two views, laying the base for setting up the stereo matching problem. In the latter section, the chapter recalls the most common hypotheses for simplifying the problem and then summarizes the fundamental steps involved in stereo matching algorithms. Next, a survey of the state-of-the-art disparity estimation methods is provided. The chapter provides the relationship between an object point in the 3D space and its 2D corresponding projections onto the image planes. Controlled Vocabulary Terms three-dimensional displays