Disparity Error in Advanced Vision Sensors

This paper reports research in which feature-based disparity error is reported against corresponding features identified in a subset of images in the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) stereo 2012 benchmark. The dataset provides real-world imagery. The theory behind finding corners or features, matching them across image pairs, and then calculating the resultant disparity is explained. These disparity results are compared against disparity data provided by the KITTI stereo 2012 dataset, to produce disparity errors. The disparity errors are graphed. Concluding remarks are discussed.