Obtaining surface normals of 3D objects using photometric stereo techniques

Abstract

In this age of modern technology, 3D imaging is extremely prevalent in many industries, and it is highly beneficial to understand more about this process. This project delves into a concept known as photometric stereo. Photometric stereo is a technique of surface normal recovery based on the surface property that the amount of light reflected off a surface is dependent on the orientation of the surface with respect to both the observer and the light source. There is much value behind this as it allows the reconstruction of a 3D surface by taking multiple images of it from a fixed viewpoint with varying lighting conditions, performing computations and obtaining the surface normals at the pixel level. This renders the object into a digitized version of itself, which can then be used in various applications. This project tackles the problem of understanding exactly how surface normals of 3D objects are obtained using photometric stereo as well as exploring its implementation in a programming language. It also addresses the limitations of conventional photometric stereo and proposes solutions based on other research works in the field.