Depth map generation based on depth from focus

The key step in 3D data generation for future 3D display system is the calculation of depth map which is a gray-level image of exactly the same size of the original 2D image data that indicates the relative distance of each pixel from the camera to the object in real world. This paper presents a practical approach to generate depth map from a series of images taken in different focuses of the same scene. This algorithm requires only Gaussian filters with some addition/subtraction functional blocks, which is very easy to implement, and the result turns out to be robust to any content of scene. A complete system from image capturing by an auto-focusing camera to depth map generation and the final 3D data rendering for three different types of 3D displays is built.