Image-Guided Depth Upsampling via Hessian and TV Priors

TL;DR

This paper introduces a novel depth upsampling method that combines sparse LiDAR data with intensity images, leveraging Hessian and TV priors to produce dense high-resolution depth maps efficiently.

Contribution

It presents a new convex optimization approach using Hessian and TV priors, solved with an ADMM algorithm, for improved depth upsampling from sparse measurements.

Findings

Achieves comparable or better depth reconstruction than existing methods.

Efficiently combines LiDAR and image data for dense depth estimation.

Validated on SYNTHIA and KITTI datasets.

Abstract

We propose a method that combines sparse depth (LiDAR) measurements with an intensity image and to produce a dense high-resolution depth image. As there are few, but accurate, depth measurements from the scene, our method infers the remaining depth values by incorporating information from the intensity image, namely the magnitudes and directions of the identified edges, and by assuming that the scene is composed mostly of flat surfaces. Such inference is achieved by solving a convex optimisation problem with properly weighted regularisers that are based on the `1-norm (specifically, on total variation). We solve the resulting problem with a computationally efficient ADMM-based algorithm. Using the SYNTHIA and KITTI datasets, our experiments show that the proposed method achieves a depth reconstruction performance comparable to or better than other model-based methods.