Face and marker detection using Gabor frames on GPUs

•Face detection and recognition algorithms.•Direct volume reconstruction and 3-D image rendering.•Image reconstruction from incomplete projection data.•Researches for parallel processing system of medical images.

Many image analysis systems such as digital image-based face authentication, reconstruction or entertainment systems, etc. require fast and accurate face detection systems. We propose in this paper, a semi-automatic procedure for computer-aided diagnosis of face marker points using Gabor frames implemented on Graphics Processing Units (GPUs). Using parallel GPU architecture in the crucial parts of the algorithm, we significantly reduce the time needed for the detection. Our approach demonstrates also the advantages of the GPUs over the CPUs in face detection, object detection and marker detection algorithms.