I am currently a postdoctoral researcher in Samsung Research America (SRA) MPI lab. I received my B.E. from Computer Science at Zhejiang University in 2012. I got my Ph.D. degree from ECE at National University of Singapore (NUS) in November 2017. My research interests include Image/Video Stitching, Video Stabilization, 3D Vision, and 3D Modeling).
Seamless Video Stitching from Hand-held Camera Inputs
SEAGULL: Seam-guided Local Alignment for Parallax-tolerant Image Stitching
Direct Photometric Alignment by Mesh Deformation
I received my Bachelor of Engineering degree in Digital Media & Technology at Zhejiang University in 2012 (GPA: 3.59/4.0). During the last two years in ZJU, I joined Dr. Hongxin Zhang's team in the State Key Laboratory of CAD&CG and worked on topics about rule-based 3D city and tree modeling.
I am currently a last year PhD candidate supervised by Cheong Loong-Fah in the department of Electrical & Computer Engineering at National University of Singa- pore. Research interests include Computer Vision (3D reconstruction, visual SLAM) and Computer Graphics (3D modeling, image stitching, video editing).
Images/videos captured by portable devices (e.g., cellphones, DV cameras) often have limited fields of view. Image stitching, also referred to as mosaics or panorama, can produce a wide angle image by compositing several photographs together. Although various methods have been developed for image stitching in recent years, few works address the video stitching problem. In this paper, we present the first system to stitch videos captured by hand-held cameras. We first recover the 3D camera paths and a sparse set of 3D scene points using CoSLAM system, and densely reconstruct the 3D scene in the overlapping regions . . .
Image stitching with large parallax is a challenging problem. Global alignment usually introduces noticeable artifacts. A com- mon strategy is to perform partial alignment to facilitate the search for a good seam for stitching. Different from existing approaches where the seam estimation process is performed sequentially after alignment, we explicitly use the estimated seam to guide the process of optimizing lo- cal alignment so that the seam quality gets improved over each iteration. Furthermore, a novel structure-preserving warping method is introduced to preserve salient curve and line structures during the warping . . .
The choice of motion models is vital in applications like image/video stitching and video stabilization. Conventional methods explored different approaches ranging from sim- ple global parametric models to complex per-pixel optical flow. Mesh-based warping methods achieve a good bal- ance between computational complexity and model flexibil- ity. However, they typically require high quality feature cor- respondences and suffer from mismatches and low-textured image content. In this paper, we propose a mesh-based pho- tometric alignment method that minimizes pixel intensity difference instead of Euclidean distance of known feature correspondences . . .
I worked as a research engineer in Dr. Tan Ping’s group at National University of Singapore for one year. During this period, I was working on projects about 3D modeling and hand tracking.
I worded as a part-time intern in the group of VMA - Visual Modeling and An- alytics of Dynamic Environments for the Mass at Advanced Digital Sciences Center (ADSC), Singapore. I was working on projects about image stitch- ing, video stabilization, and motion segmentation under the supervision of Dr. Jiang Nianjuan.