The reconstruction of object surfaces is an important area in computer vision. This area is directed towards the recovery of three-dimensional information of an object placed in a scene. Most classical existing reconstruction methods assume that objects in the scene reflect light equally in all directions. This assumption on surface reflectance is referred as a Lambertian BRDF (Bidirectional Reflectance Distribution Function) however, Lambertian reflectance is not fulfilled for real-world objects, leading to incorrect depth estimates. Several methods propose a partial solution taking advantage of calibrated cameras and lights or some approximation of the surface reflectance. Reflectance fields are a common image-based model used for rendering the appearance of objects with a high-degree of realism. Reflectance fields are used in a variety of applications. For example, they are used to capture the appearance of real-world objects with complex geometry, like human bodies, furry teddy bears, or bonsai trees. They are also used to represent intricate distributions of light, like the illumination from a flash light. However, despite the increasing popularity of using reflectance fields, fast methods do not exist for acquiring them. This thesis presents two main contributions towards reflectance field acquisition and three-dimensional reconstruction. The first is a method for acquiring efficiently the reflectance field data to provide the surface reflectance information. This method introduces the use of independent component analysis, it enables fast acquisition of the reflectance field. The prototype acquisition system consists of a pair of projector and camera controlled by a computer. The effectiveness of the system is demonstrated with scenes composed of different material properties where the reflectance quality is maintained. The second contribution is a three-dimensional reconstruction of real-world objects placed in the scene. Three-dimensional reconstruction is obtained by relating depth estimations with reflectance fields. The reflectance field is integrated with threedimensional reconstruction equations by considering the system setup as a stereo system, it provides depth estimations of objects exhibiting different material properties with high accuracy.