Chapter 31 – Real-Time Stereo on GPGPU Using Progressive Multiresolution Adaptive Windows

Publisher Summary This chapter presents a new GPGPU-based real-time dense stereo-matching algorithm. The algorithm is based on a progressive multiresolution pipeline that includes background modeling and dense matching with adaptive windows. For applications in which only moving objects are of interest, this approach effectively reduces the overall computation cost quite significantly and preserves the high definition details. Estimating depth from stereo is a classic computer vision problem, which has received tremendous attention since the early days. Recovering 3D information from a pair of stereo cameras has been a popular topic because the additional 3D information provided by this technology contains significantly more information than 2D information produced by traditional cameras. Some believe that this technology will fundamentally revolutionize the computer vision signal-processing pipeline, as well as how future cameras will be built. The current implementation achieves 36 Hz stereo matching on 1024 × 768 stereo video with a fine 256-pixel disparity range. The focus of this work is to provide efficient high-resolution stereo algorithms for real-time applications in which only foreground moving objects are of interest, such as motion capture, object tracking, and recognition and identification in a surveillance scenario.