Real-time Correction of Images for Stereo Matching: Real-time Rectification of Images Based on Customized Hardware

Stereo vision is one of the most effective methods used in the field of machine visionary because of its high measurement accuracy compared to other techniques. This mimics the biological vision system where streams of image pairs are captured by the human eyes and processed by the brain in order to reconstruct a 3D realization of the surrounding environment. This method is based on matching the corresponding points in two images in order to generate 3D depth maps of the environment scenes. Applying stereo matching in computer hardware, however, is challenging because it is computationally expensive and requires real-time processing of high resolution images; as the case in autonomous vehicles. This book outlines a new proposed method to rectify high resolution images in real-time based on high speed data streaming protocol, which is implemented on FPGA. This method is still under research and development by Dr. John Morris and his research group in UoA for further refinement.