Holoimages

We introduce a novel geometric representation called the holoimage, which encodes both shading and geometry information within the same image, based on the principles of wave optics. 'Image' referes to the representation and records the amplitude of the lighting; 'holo' means that it encodes phase, and hence, three-dimensional information. Compared to conventional geometry images or depth images, the holoimage has much higher geometric accuracy. Thus, 3D information can readily be stored and transmitted using the common 24-bit image format. Holoimages can be efficiently rendered by modern graphics hardware; rendering speed is independent of the geometric complexity and only determined by the image resolution. Rendering holoimages requires no meshes, only textures. Holoimages allow various geometric processing tasks to be performed simply using straightforward image processing methods, including such tasks such as embossing and engraving, geometric texture extraction, and surface deformation measurement. Conventional geometric representations, such as meshes, point clouds, implicit surfaces and CSG models, can be easily converted to holoimages using conventional rendering techniques in real time. The opposite process, converting holoimages to geometry in the form of a depth map is accomplished efficiently accomplished by graphics hardware. Furthermore, holoimages can be easily captured from the real world with a projector and a camera at video frame rate.