Authors

Presenter(s)

Alison Hardie

Comments

3:00-4:15, Kennedy Union Ballroom

Files

Download

Download Project (1.8 MB)

Description

LAser Detection And Ranging (LADAR) is widely used in fields such as forestry, topographic mapping, autonomous driving, urban planning, robotics, and object recognition. Automated tools are needed to label and process this data, as manual labeling is tedious and time consuming. Region growing is a widely used technique for both 2D and 3D segmentation. In seeded region growing, segmentation begins at a seed point, and similar neighbors are iteratively added to the region. This approach is applied here using superpoints generated by SuperPoint Transformer (SPT). The use of superpoints improves processing efficiency and captures features on a larger scale. In this method, the user clicks on an object to select a seed point. Geometric features are used to define a similarity metric which guides the iterative region expansion, including neighboring superpoints that meet the similarity criteria. This approach enhances LADAR segmentation and labeling, making the process more efficient and scalable.

Publication Date

4-23-2025

Project Designation

Graduate Research

Primary Advisor

Vijayan K. Asari, Theus H. Aspiras

Primary Advisor's Department

Electrical and Computer Engineering

Keywords

Stander Symposium, School of Engineering

Institutional Learning Goals

Scholarship

Superpoint-Based Region Growing for Point Cloud Labeling

Share

COinS