A Survey of Photometric Stereo Techniques

Abstract

Reconstructing the shape of an object from images is an important problem in computer vision that has led to a variety of solution strategies. This survey covers photometric stereo, i.e., techniques that exploit the observed intensity variations caused by illumination changes to recover the orientation of the surface. In the most basic setting, a diffuse surface is illuminated from at least three directions and captured with a static camera. Under some conditions, this allows to recover per-pixel surface normals. Modern approaches generalize photometric stereo in various ways, e.g., relaxing constraints on lighting, surface reflectance and camera placement or creating different types of local surface estimates.Starting with an introduction for readers unfamiliar with the subject, we discuss the foundations of this field of research. We then summarize important trends and developments that emerged in the last three decades. We put a focus on approaches with the potential to be applied in a broad range of scenarios. This implies, e.g., simple capture setups, relaxed model assumptions, and increased robustness requirements. The goal of this review is to provide an overview of the diverse concepts and ideas on the way towards more general techniques than traditional photometric stereo.

State‐of‐the‐art in Multi‐Light Image Collections for Surface Visualization and Analysis

“…The most straightforward way, and classic way, to extract a single image that conveys object surface information from a MLIC is the albedo visualization. This is a light‐independent view of the low‐frequency (typically constant), spatially‐varying BRDF across the surface [CDG∗18, Woo80, AG15]. Since the albedo removes highlights or other high‐frequency effects that might disturb the underlying content in the visualization, this approach is often used for objects where pigment information is important (e.g., think of a painting inspection).…”

“…However, this minimal setup leads to undersampling of the reflectance function and missing data due to shadows, so most of the MLIC processing is done with a much larger number of lights uniformly sampling the entire sphere or hemisphere around the scene. Implementations of this approach include large static light‐domes approximating a far‐light condition [MGW01, PSM05, CHI19, HP15, EBB∗11, AIK13], smaller micro‐domes [EBB∗11, FBKR17, Ham15, WVM∗05, PBFS14, VVP∗18, HBMG02] requiring the handling of non‐collimated light‐rays through near‐light or spot‐light models [AG15, PCGG16, HWBC15], or virtual domes made with moving light arcs or robotic arms with a small number of light sources to span the entire hemisphere [MBW∗14, SOSI03, DCCS06]. The known regular arrangement of lights in these solutions, using tens to hundreds of lights, allows methods, user interfaces, and tools to exploit a parameterization of the light space for navigation and/or interpolation (Sec.…”

Photometric stereo with only two images: A theoretical study and numerical resolution

“…13, which are available on the web 4 , show a plaster bust of Beethoven illuminated by three non-coplanar, parallel and uniform light beams. Since the light vectors are provided, these real data are particularly well adapted to the PS3 technique (see Section 2.2).…”

“…Among all the photographic 3D-reconstruction techniques, we focus in this work on shape-fromshading (SFS) and photometric stereo (PS), which exploit shading information when one (SFS) or several (PS) sources illuminate the observed object. For a comprehensive overview on these techniques, see the reference book [1] by Horn and Brooks, but also [2] and [3,4] for up-to-date surveys on SFS and PS, respectively.…”

Potential and Limitations of Photometric Reconstruction Through a Flock of Dove Cubesats

“…Current research is focused on subjects with non-Lambertian reflection and/or unknown lighting (Ackermann and Goesele, 2015). In Earth observation the ill-posed formulation of photometry, photoclinometry, has been applied (Peng et al, 2015).…”