Adaptive noise reduction method using noise modeling for TOF sensor

Abstract

In this paper, we propose an advanced way of reducing depth noise for time-of-flight (ToF) camera. Most of depth denoising method uses a filter with fixed parameter or sets parameter considering one factor: amplitude or distance. However, the depth noise is dependent on both amplitude and distance value. Thus, the characteristic of depth noise should consider two factors simultaneously: distance and amplitude. In proposed method, the parameter of adaptive bilateral filter is set to value from our noise model using distance and amplitude information of pixel. In experiments, the proposed method decreases depth noise more effectively while preserving details than normal bilateral filter with fixed parameters.