Zhidong Liang

Papers

Total Citations

H-Index

About

Zhidong Liang is a leading researcher in 3D computer vision and robotics, with a primary focus on point cloud registration—a critical challenge for autonomous navigation, scene understanding, and augmented reality. His most influential work, "CentroidReg: A Global-to-Local Framework for Partial Point Cloud Registration" (2021, 20 citations), addresses a fundamental limitation in existing registration methods: global algorithms are brittle under noise and partial occlusion, while local methods rely heavily on accurate initial alignment. Liang’s proposed framework elegantly bridges this gap by first estimating global centroids to provide robust coarse alignment, then refining locally for precision. This hybrid approach significantly improves registration accuracy in challenging real-world scenarios, such as cluttered or partially overlapping scans. Beyond this flagship paper, Liang’s research portfolio consistently advances robust geometric perception, contributing to more reliable 3D mapping and object recognition systems. His work has been recognized for its practical impact, offering a principled solution that balances efficiency and robustness. For students and researchers entering the field, Liang’s contributions exemplify how thoughtful algorithmic design can overcome persistent bottlenecks in 3D data processing, making his papers essential reading for anyone working on point cloud-based perception.