Su-Yong An

Dr. Su-Yong An

orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 February 2022 in IEEE Robotics and Automation Letters
Verified
Journal Article
Open access
SASu-Yong An
Su-Yong An
JKJaeyoung Kim
Jaeyoung Kim
Global localization (or place recognition) is a method of finding the current location of a robot on a map generated by a mapping process, and it is an open field that has not yet been completely solved in the field of mobile robotics. Most existing approaches to global localization are based on extraction of interest point features and their descriptors whether from raw laser scans or 2D occupancy grid maps. In this letter, unlike most approaches, we propose a novel method of extracting a statistical signature of geometric and structural features from a submap. A boundary and free-space features can characterize a geometric shape, while a reflection symmetry can quantify a structural shape of the submap. Experiments using five pre-built map publicly available demonstrate that the proposed method outperforms the other state-of-the-art image-based methods by examining precision-recall curve especially when occupancy noise added to the submap is progressively increased.

1 December 2010
Verified
Conference Paper
Closed access
SASu-Yong An
Su-Yong An
JKJeong-Gwan Kang
Jeong-Gwan Kang
LLLae-Kyoung Lee
Lae-Kyoung Lee
SOSe-Young Oh
Se-Young Oh
This paper presents a simultaneous localization and mapping (SLAM) of a large indoor environment using Rao-Blackwellized particle filter (RBPF) along with line segments as the landmarks. To represent the environment as a compact form, we use only two end points of the line segment, reducing computational cost in modeling line uncertainty. With a modified scan point clustering method, the proposed adaptive iterative end point fitting (IEPF) plays an important role in estimating line parameters by taking a noisy scan point near end points into account. Thus, by line-segment matching the robot is localized well in a local frame. We also introduce an online global optimization of a map, which provides more consistent map by removing spurious lines and merging collinear lines. Each of our approaches is efficiently integrated into the proposed RBPF-SLAM framework. Experiments with well-known data set demonstrate that the proposed method provides a reliable SLAM performance along with a compact map representation.

16 July 2015 in Autonomous Robots
Verified
Journal Article
Closed access
PDF file
SASu-Yong An
Su-Yong An
LLLae-Kyoung Lee
Lae-Kyoung Lee
SOSe-Young Oh
Se-Young Oh

1 August 2011
Verified
Conference Paper
Closed access
LLLae-Kyoung Lee
Lae-Kyoung Lee
SASu-Yong An
Su-Yong An
SOSe-Young Oh
Se-Young Oh
This paper presents a new approach to solve the problem of real-time robust face detection and tracking in complex environment. We propose a two level approach to detect faces and tracking. The lower level of the approach implements the extraction of face candidates using the combination of skin color model and haar-like features based adaboost learning algorithm. With this method, multiple-view faces are able to be detected in real-time with high recognition accuracy. The higher level of approach implements the robust face tracking with extended CAMSHIFT (Continuous Adaptive Mean SHIFT). The experimental results show that the proposed algorithm is robust and efficient to detect and track the face-of-interest in the cases of clutter background and the occurrence of occlusion.

18 August 2025 in Remote Sensing
Verified
Journal Article
Open access
PDF file
WKWonjune Kim
Wonjune Kim
HSHyun-Sik Son
Hyun-Sik Son
SASu-Yong An
Su-Yong An
Accurate estimation of tree diameter at breast height (DBH) from LiDAR point clouds is essential for forest inventory, biomass assessment, and ecological monitoring. This paper presents a perimeter-based DBH estimation framework that achieves competitive accuracy against geometric fitting methods across three datasets. The proposed approach partitions the trunk cross-section into angular sectors and employs Gaussian Mixture Models (GMMs) to identify representative boundary points in each sector, weighted by radial proximity and statistical confidence. To handle occlusion and partial scans, missing sectors are reconstructed using symmetry-aware proxy generation. The final perimeter is modeled via either convex hull or B-spline interpolation, from which DBH is derived. Extensive experiments were conducted on two public TreeScope datasets and a custom mobile LiDAR dataset. Compared to the Density-Based Clustering Ring Extraction (DBCRE) baseline, our method reduced RMSE by 22.7% on UCM-0523M (from 2.60 to 2.01 cm), 34.3% on VAT-0723M (from 3.50 to 2.30 cm), and 29.6% on the Custom Dataset (from 2.16 to 1.52 cm). Ablation studies confirmed the individual and synergistic contributions of GMM clustering, radial consistency filtering, and proxy synthesis. Overall, the method provides a flexible alternative that reduces dependence on strict geometric assumptions, offering improved DBH estimation performance with moderate occlusion and incomplete, uneven boundary coverage.

17 May 2025 in arXiv
Verified
Preprint
Open access
PDF file
WKWonjune Kim
Wonjune Kim
LLLae-Kyoung Lee
Lae-Kyoung Lee
SASu-Yong An
Su-Yong An
We report on the application of a high-capacity semantic segmentation pipeline to the GOOSE 2D Semantic Segmentation Challenge for unstructured off-road environments. Using a FlashInternImage-B backbone together with a UPerNet decoder, we adapt established techniques, rather than designing new ones, to the distinctive conditions of off-road scenes. Our training recipe couples strong photometric distortion augmentation (to emulate the wide lighting variations of outdoor terrain) with an Exponential Moving Average (EMA) of weights for better generalization. Using only the GOOSE training dataset, we achieve 88.8\% mIoU on the validation set.