Research in IFP Group

The faculty in IFP Group are concerned with a wide spectrum of research issues related to the formation and manipulation of images. The many research topics under study fall into three broad categories: computerized imaging; image/video transmission, storage, and manipulation; and image and scene modeling and analysis.

Computerized Imaging

Computerized imaging is concerned with the signal processing techniques and algorithms required to form 2D or 3D images from multiple sensed data sets. Important examples include x-ray tomography, magnetic resonance imaging, and synthetic aperture radar. The objective is to obtain images of high quality from data which may be noisy or incomplete. Applications of this research include improved image formation in medical and industrial CT and MRI scan instruments, higher quality images of the ground terrain from airborne or spaceborne radar, and automotive radar for use under conditions of poor optical visibility.

Image/Video Transmission, Storage, and Manipulation

In image/video transmission, storage, and manipulation, the key issue is Representation. One strives to find representations of images and video which are efficient (requiring a small amount of bits to transmit or store) yet easy to manipulate (e.g., easy to find what one is looking for in a database). The group carries out research in image and video compression using a variety of approaches ranging from wavelets and fractals, to ideas from pattern recognition and computer vision. Also under study are methodologies and techniques for image/video indexing and editing. Research in human visual perception is an essential part of this effort since, in many applications, the images after processing are viewed by humans, and the subjective quality of images is an important performance criterion. Applications of this research include video phone, teleconferencing, and multimedia databases.

Image and Scene Modeling and Analysis

In image and scene modeling and analysis, a major research effort is the modeling, analysis, and visualization of human facial movement and hand gestures. The approach is to first construct good 3D models of human head/face and hand from multiple sensed 2D images, and then use the models to do analysis and synthesis. Applications include vision-based human computer interfaces, very low bitrate model-based video compression for video phone and teleconferencing, and computer recognition of American Sign Language. Another research effort is the detection and tracking of coherent patterns in turbulent fluid flow. The aim is to develop tools of image analysis which can help both scientists in their basic research on fluid dynamics and engineers in their design of prototypes of turbomachinery and vehicles. Among the three main research themes of the Beckman Institute, the research of the IFP Group is related most closely to Human-Computer Intelligent Interaction (HCII). In a broad sense, HCII means the synergistic collaboration of human intelligence and machine intelligence in problem solving. Vision and image based techniques will clearly play a key role in most HCII scenarios. A major scenario of interest to the IFP Group is collaborative manufacturing product prototyping in a virtual environment (VE). Here vision/image based techniques can potentially make the interaction between people and the VE more natural and more efficient.