Edge Artifact Correction for Doppler Time-of-Flight Radial Velocity Imaging Based on Adaptive Gradient Threshold

Doppler time-of-flight (Do-ToF) imaging has recently attracted significant attention due to its high-resolution capabilities for measuring radial velocity. However, a challenge arises when the back-reflected signal received by pixels switches between a moving object and a stationary background, leading to the appearance of edge artifacts in the velocity images. To address this issue, we propose a per-pixel gradient-based method for identifying and correcting these artifacts. First, the homodyne measurement was modelled to acquire the gradient feature of regular pixels and those affected by edge artifacts. Building on this, we proposed a self-adaptive gradient threshold method to accurately identify artifacts, thereby addressing misidentification caused by sinusoidal patterns in the moving objects' measurements. The identification method was implemented with five continuous results to determine edge artifacts. The linear prediction was referred to correct the artifact region. Experimental results demonstrate that the proposed method effectively reduces edge artifacts. The average velocity error was reduced by 78.5% compared to the original results. The root mean square error (RMSE) was reduced by 62.7%. The effectiveness of the adaptive gradient threshold was validated, with the RMSE reducing by 31.1% compared to the preset gradient threshold.