Intrinsic Image Decomposition

Intrinsic Image Decomposition is the process of breaking down an image into its constituent components, including reflectance (albedo) and illumination. Reflectance represents the color of the object, independent of the camera viewpoint and lighting conditions; while illumination depends on the camera viewpoint and object geometry, encompassing effects such as shadows, lighting, and interreflections. This technique is beneficial for many computer vision algorithms, such as shape from shading recovery algorithms that can leverage important visual cues in the illumination image, and segmentation and detection algorithms that benefit from the reflectance image being free of complex lighting influences. Additionally, intrinsic images have a wide range of applications in computational photography, involving material recoloring, relighting, texture reassignment, and stylization.