Remove Projective LiDAR Depthmap Artifacts via Exploiting Epipolar Geometry

Shengjie Zhu*, Girish Chandar Ganesan, Abhinav Kumar, Xiaoming Liu ;

"sensing is a fundamental task for Autonomous Vehicles. Its deployment often relies on aligned RGB cameras and . Despite meticulous synchronization and calibration, systematic misalignment persists in projected . This is due to the physical baseline distance between the two sensors. The artifact is often reflected as background incorrectly projected onto the foreground, such as cars and pedestrians. The dataset uses stereo cameras as a heuristic solution to remove artifacts. However most AV datasets, including , , and , lack stereo images, making the solution inapplicable. We propose , a parameter-free analytical solution to remove the projective artifacts. We construct a binocular vision system between a hypothesized virtual camera and the RGB camera. We then remove the projective artifacts by determining the epipolar occlusion with the proposed analytical solution. We show unanimous improvement in the State-of-The-Art () monocular depth estimators and object detectors with the artifacts-free ."