Partitioned iterated function systems by regression models for head pose estimation

Abstract Head pose estimation represents an important computer vision technique in different contexts where image acquisition cannot be controlled by operator, making face recognition of unknown subjects more accurate and efficient. In this work, starting from partitioned iterated function systems to identify the pose, regression models are adopted predict angular value errors (yaw, pitch roll axes, respectively). This method combines fractal compression characteristics, such as self-similar structures order similar head rotation, with analysis prediction. The experimental evaluation is performed on widely used benchmark datasets, i.e., Biwi AFLW2000, results compared many existing state-of-the-art methods, demonstrating robustness proposed fusion approach excellent performance.