Adaptive Detection and Elimination Method Based on Phase Resolution for Motion Blur in ITOF Sensors

The indirect time-of-flight sensors acquire the depth map by determining the phase offset between the emitted and reflected infrared light. Motion blur may occur in the depth image if the positions of objects within the scene change during the data acquisition process. A method for detecting motion blur that is based on phase resolution is proposed in this paper. The threshold for detecting motion blur is adaptively established using the output data from the sensor. Then, a depth threshold is established to differentiate between the depth of non-blurred pixels and categorize them into object depth and background depth. For pixels with motion blur, they are classified into two types based on the object’s motion trajectory: background-to-object pixels and object-to-background pixels. These blurred pixels are then replaced with object depth and background depth. Experiments on data generated by the Mitsuba Renderer demonstrated that the proposed method effectively eliminates motion blur, achieving mean depth error reductions of 37.8%, 79.3%, and 67.1% for moving objects at speeds of 0.3 m/s, 0.5 m/s, and 1 m/s, respectively. Furthermore, due to its low computational complexity, the method requires only 14.96 milliseconds to process depth images with a 512 x 512 pixels resolution. Consequently, this method is well-suited for real-time depth image processing.