An effective phase-unwrapping method for time-of-flight camera based on 2D-Gaussian model

The time-of-flight camera produces a depth map using the phase delay between the emitted infrared (IR) light signal and the received one. However, for any objects farther than a certain range, called the non-ambiguity range (NAR), the measured depth value is much smaller than its actual value. To overcome the problem, this paper presents a novel method to effectively correct the depth value of the region with the depth ambiguity (RDA). In the proposed method, the amplitude map of the reflected IR signal on the scene points is modeled as a 2D-Gaussian function. Based on this model, we extract the RDA from the original depth map and then add the NAR to the depth value of the RDA for depth correction. Experimental results show that the proposed method effectively reduces the RDA compared with conventional ones.