Synthetically Generating Motion Blur in a Depth Map from Time-of-Flight Sensors

Abstract

Motion blur is a common artifact that affects imaging systems. Synthetically creating motion blur in two-dimensional (2D) images is a well-understood process and has been used to develop deblurring systems. However, there are no well-established techniques for synthetically generating motion blur within three-dimensional (3D) images since the behavior of motion blur within 3D images, such as depth maps and point clouds, is not as well-understood. In this work, we develop a simple and accurate framework to synthetically generate motion blur within depth maps that accurately captures the behavior of the real motion blur that is encountered using a Time-of-Flight (ToF) sensor. We develop a probabilistic model that predicts the location of invalid pixels that are typically present within depth maps that contain real motion blur. We also introduce a method to quantify the performance of a synthetic motion blur filter for depth maps based on a comparison between a depth map with synthetic motion blur and a depth map with real motion blur. Our results indicate that our framework is able to achieve an average Boundary F1 (BF) score of 0.8864 for invalid pixels for synthetic radial motion blur and a BF score of 0.8401 for synthetic linear motion blur.