1. Introduction
Accurate geometric information is vital for a wide range of applications in image processing and computer vision, including geometric reconstruction [32] and visual odometry [25]. However, depth data acquired from sensors inevitably contains noise and missing values due to physical reasons. The severity of these disturbances is influenced by factors such as the distance to the object and coordinates of the pixel [29]. As a consequence, these effects can substantially undermine the performance of algorithms that heavily rely on dependable geometric information.