Current Location: Home» Research Field»

Research field

The Lab research focuses on fundamental theory and engineering technology about computer graphics and visualization. Currently, our research interests include:

  • Intelligent processing on internet images and videos
  • Non-photorealistic rendering and stylization
  • Multimodal image based 3D reconstruction
___________________________________________________________________________________________________

Intelligent processing on internet images and videos

The development of internet technology and prevalence of digital cameras make it easy to acquire and share images and videos, which results in massive multimodal and heterogeneous big data resource. However, the rapid increase on data amount cannot promise high quality visual effect, which affects many applications on community and industry. Hence, intelligent processing on internet images and videos brings new challenges as well as opportunities for computer graphics.

The lab primarily works on the following 3D reconstruction techniques:
  1. Image and video enhancement techniques
  2. Interactive image and video editing and composition
  3. Large scale images and videos exploration
  4. Image and video quality assessment
___________________________________________________________________________________________________

Non-photorealistic rendering and stylization

Non-photorealistic rendering is to produce images and videos with artistic effect by analyzing and processing the contents. With the prevalence of digital media, rapid and intelligent artistic rendering on images and videos enables a broad range of applications. The popular artistic styles include cartoons, pencil sketch, oil painting, watercolor painting, etc.

The lab primarily works on the following 3D reconstruction techniques:
  1. Content-aware non-photorealistic rendering
  2. Filtering based image and video stylization
  3. Image stylization on portrait
___________________________________________________________________________________________________

Multimodal image based 3D reconstruction

Research on 3D reconstruction and its geometry processing is one of the hottest topics in computer graphics. 3D reconstruction refers to computational techniques that recover 3D information (like shape) from 2D projection. The reconstructed 3D model can be observed from arbitrary views in the space, with 3D drawing effect. The trend of 3D reconstruction techniques involves large scale of data, multimodal fusion, high precision and speed. Geometry processing is to further perform semantic analysis for high quality modeling and rendering.

The lab primarily works on the following 3D reconstruction techniques:
  1. Large scale scene reconstruction based on aerial videos
  2. Indoor reconstruction based on RGB-D camera
  3. Human body reconstruction based on depth and video fusion