Appearance Modelling and Rendering
We are interested in the acquisition and modeling of the complex phenomena of object appearance. We are researching realistic image synthesis algorithms, with emphasis for rendering large models. The tremendous resolution and the high computational complexity of these image synthesis tasks present big challenges to modeling, processing, interaction at interactive rates.
We are interested in the simulation of Chinese ink and wash painting technologies. We not only study the algorithm, but also develop the interactive and real time software system .
In the field of archaeology, the traditional way of drawing and sketching involves intensive manual labors and rich specialist skills. We are interested in the development of efficient (semi)automatic techniques to help archeologists in the drawing or sketching, including automatically unroll the decorations on 3D pottery vessels into 2D planar space, (semi)automatically generate sketch or sketch recommendation.
Multiview 3-D reconstruction
We are interested in the design of multi-view multi-light capture system, including the geometric calibration and color calibration methods,, 3-D reconstruction of the dynamic scene via low-frame rate camera.
Large Scale Data Visualization and Visual Analytics
Cultural Heritage Data Visualization
We investigate the use of data analysis and exploration with novel visualizations and expand visualization study into cultural heritage domain. Current emphasis is on the analysis of relics degradations, the risks the relics faced. By the studying of the correlation between the relics and the risks and the environments, we help the curators of the heritage sites transform the conservation of cultural heritages from passive rescuing conservation to proactive preventive conservation.
We are developing scalable systems for interactive visual exploration and analysis of problems of modern metropolis faced everyday, with emphasis on public utility problems(water, gas, drainage, traffic). We research efficient techniques to combine visualization and domain knowledge together, involve the domain experts in the analysis loop, to allow the administrators from different departments to interactively explore and analyze their urban data.
We are researching in-situ visualization methods for large-scale simulation in high performance computing, with emphasis on visualization of in-situ pipeline optimization, the efficient and scale data i/o, data compression algorithms and feature extraction methods.
Color and depth Estimation and super resolution
We are interested in the image restoration of corrupted images(with noise, with haze, or with low-resolution), such as image denoise, deblocking, dehaze and super resolution.
Panorama Image Processing
We are interested in the panorama image processing, with emphasis on spherical image feature description, spherical image stitching and spherical image editing.
Image Relocalization and Relighting
We are interested in accurately relocate target objects and measure the difference between two shots. Apparently the emphasis is on how to accurate image registration, active 6 DOF spatial estimation and control, and lighting estimation and control.