Absolute reflectivity is measured in photometric stereo by using at least three images with different illumination directions where each image produces a conic curve in gradient space. These curves intersect at points representing possible surface orientations. The additional third illumination resolves any remaining ambiguities and quantifies absolute surface reflectance based on intersection congruence. Accurate reflectivity measurement allows for verification of the Lambertian surface model and enables precise surface characterization essential for accurate 3D modeling .

Photometric stereo enhances surface analysis by using images illuminated from multiple light sources, reducing possible gradient positions in surface orientation. With each illumination, surface orientation is narrowed down to specific intersections in gradient space, reducing ambiguity regarding surface gradients. Employing three light sources resolves the final ambiguity in these orientations, provides data on absolute surface reflectance, and tests the Lambertian surface assumption. This method increases accuracy in determining surface characteristics compared to monocular imaging, which lacks such triangulation capabilities .

Lambertian surfaces are significant in computer vision as they provide a simple model where reflectance depends only on the angle of incidence, ideal for many lighting analysis techniques like photometric stereo. The reflectance function's independence from emergence or phase angles simplifies deriving gradients for surface orientation. Such simplicity aids in reconstructing accurate 3D models by allowing consistent assumptions across varying illumination conditions. Thus, Lambertian models serve as a benchmark when testing more complex reflection behaviors in real-world applications .

In perspective projection, vanishing points arise where lines, known to be parallel in the physical world, appear to converge within an image. This occurs as rays from these lines meet at a focal point such as a camera lens. A vanishing point on the horizon line is a specific instance where all parallel lines on a ground plane meet. Vanishing lines, another concept in perspective projection, represent a plane's perspective with its own potential for vanishing points. These concepts demonstrate how real-world geometry is projected with depth cues onto a 2D plane .

Shape from Shading techniques model the reflectance of surfaces using angles of incidence, emergence, and phase. For Lambertian surfaces, which reflect light uniformly, the reflectance function depends solely on the incidence angle. By considering these parameters, a reflectance map is constructed in gradient space representing contours of equal brightness. When the light source direction aligns with the viewing direction, these contours depict possible surface orientations. The technique is most effective when specular highlights are minimal, typical in non-shiny surfaces .

Texture aids depth perception by providing cues through the deformation of textural elements. As demonstrated by Ohta et al., circular patches on surfaces project as ellipses and appear progressively flatter with increased obliqueness when moving into the distance. Identifying the horizon line helps in interpreting these depth cues, which involves analyzing the areas of paired texture elements. The process utilizes a Hough-based procedure to eliminate false alarms and reliably fit a line in image space for horizon detection .

Orthographic projection represents objects using parallel lines to project points onto a flat plane, typically used in engineering drawings to show multiple views (plan, side, elevation) without perspective distortion. Perspective projection, however, involves converging rays that pass through a lens, leading to scale changes and perspective distortions. This is observed when parallel lines in real life appear to converge towards a vanishing point on the horizon, a representation of infinite points on a ground plane. Therefore, while orthographic projection is a special case of perspective projection from a distant point, perspective projection involves depth cues like vanishing points and horizon lines .

Detecting horizon lines in textured images is challenging because it requires distinguishing between real horizon information and noise to accurately infer depth. The technique relies on comparing areas of texture elements to deduce potential horizon positions, accumulating data across image space. Hough-based fitting eliminates false predictions, but initial horizon line estimates must be reliable, as errors here can cascade into inaccurate depth interpretations. Therefore, resolving such challenges necessitates precise computational techniques to differentiate genuine texture-induced depth cues from artifacts .

Using different light sources in photometric stereo is necessary to resolve ambiguities in surface gradient determination. Each light source direction affects which surface orientations are highlighted in the geometric configurations of gradient space. Three independent light sources typically introduce three intersecting conic curves, whose precise intersection point yields unambiguous surface gradient data. This diversity of light directions permits more accurate deductions of surface reflectance properties and orientation, essential for detailed 3D scene reconstruction .

The 'phase' angle in Shape from Shading is used alongside the angles of incidence and emergence to model the reflectance properties of a surface. Although the phase angle is often immaterial for perfectly Lambertian surfaces, it can provide additional information about the surface's reflective characteristics when they deviate from this ideal model. Accurately incorporating the phase angle can thus refine the reflectance maps used in deducing surface orientations, enabling more comprehensive analyses of surface lighting phenomena .