Efficient LiDAR Reflectance Compression via Scanning Serialization

Reflectance attributes in LiDAR point clouds provide essential information for downstream tasks but remain underexplored in neural compression methods. To address this, we introduce SerLiC, a serialization-based neural compression framework to fully exploit the intrinsic characteristics of LiDAR reflectance. SerLiC first transforms 3D LiDAR point clouds into 1D sequences via scan-order serialization, offering a device-centric perspective for reflectance analysis. Each point is then tokenized into a contextual representation comprising its sensor scanning index, radial distance, and prior reflectance, for effective dependencies exploration. For efficient sequential modeling, Mamba is incorporated with a dual parallelization scheme, enabling simultaneous autoregressive dependency capture and fast processing. Extensive experiments demonstrate that SerLiC attains over 2× volume reduction against the original reflectance data, outperforming the state-of-the-art method by up to 22% reduction of compressed bits while using only 2% of its parameters. Moreover, a lightweight version of SerLiC achieves ≥ 10 fps (frames per second) with just 111K parameters, which is attractive for real-world applications.

The input 3D LiDAR point cloud is first serialized into 1D ordered point sequences, which are then divided into windows for parallel processing. For each window, a Mamba-driven autoregressive coding (MDAC) scheme is employed, which embeds scanning index (${\mathbf{F}}_i^{pos}$), radial distance (${\mathbf{F}}_i^{\rho }$), and prior reflectance (${\mathbf{F}}_{i-1}^{\mathbf{x}}$) as context to generate the probability mass function (PMF) for the reflectance intensity of the target ($i$-th) point.

We presents a detailed comparison of the overall bit rate and CR gains of SerLiC against G-PCC (RAHT), G-PCC (Predlift), and Unicorn.

This paper presents SerLiC, a serialization-based neural compression framework tailored for LiDAR reflectance attribute. By leveraging scan-order serialization, SerLiC transforms 3D point clouds into 1D sequences, aligning with LiDAR scanning mechanisms and enabling efficient sequential modeling. The Mamba model with physics-informed tokenization further enhances its ability to capture points correlations autoregressively, while maintaining linear-time complexity. Its high efficiency, ultra-low complexity, and strong robustness make it a practical solution for real LiDAR applications. Future work will extend SerLiC to lossy compression for higher compression efficiency while preserving essential information for downstream tasks.