Logo: University of Southern California

By Research Areas


                                                                    Research Areas

Adaptive Signal Processing

Recent work focuses on modeling the underground structure of oilfields using system identification tools as well as network tomography. The main goal in this work is to exploit this structure in order to increase the overall production rates in real oilfields. Research has also been conducted in the context of internet traffic analysis, where various detection, estimation and sampling methods have been developed to help identify certain types of network attacks (e.g., denial of service attacks).


Audio/Music Signal Processing Subgroup

Audio/music signal subgroup aims to propose digital signal processing techniques to solve problems of audio and music applications. Our research interests could be categorized as: (1)computer perception of music: music onset and beat detection, music segmentation, instrument classification, (2) multimedia security: watermark, data hiding and (3) blind source separation: the separation of speech and music, separation of different sounds in music.



Behavioral Signal Processing

The behavioral signal processing group at SAIL is interested in technology-based assessment of language learning skills, robotics and speech processing technology for autistic children, and creative technologies with applications in theatre.


Brain Imaging Subgroup

Brain imaging uses recordings of brain activity via magnetoencephalography (MEG) or electroencephalography (EEG) to determine how the brain reacts to different stimuli. Our goal is to use E/MEG to determine 1) what regions of the brain activate to a given stimulus and 2) what regions of the brain work together to form a network active (or inactive) based on a stimulus (or lack thereof).


Cardiovascular, Obesity and Vocal Tract

Our goal is to research and improve methods of heart and vascular imaging for early diagnosis and characterization of cardiovascular disease. Our interests include arterial spin labeling, 3D perfusion imaging, flow-sensitive methods, and coronary artery dilation detection. We study MRI techniques related to the assessment of obesity and potential interventional methods.



We study and analyze critical issues in advanced communication systems including CDMA, OFDM, UWB and MIMO. Novel transmitter and receiver design methodology and algorithms have been proposed to overcome fading in wireless channels. Moreover, cross-layer design has been developed to provide better resource allocation on wireless sensor networks and cooperative networks.


Computer Graphics

Our research focuses on diverse fields in computer graphics including 3D data compression, time-varying geometry, simulation and rendering of liquids and clouds, physical based human motion synthesis, Non-Photorealsitic Rendering in video game, and simulation of interaction between plants and natural forces. 


Distributed Data Compression

Current research focuses on the design of distributed transforms that can be computed in-network, with focus primarily in wireless sensor networks. These transforms de-correlate data across neighboring nodes, thereby leading to some data reduction and, ultimately, compression. In this context, wavelet transforms designed along communication graphs have been developed as well as methods which use compressed sensing.



 The areas that we are studying are emotion recognition in speech, expressive speech synthesis, multi-modal analysis of human expressions, expressive speech production, emotions in text, expressive human-robot interfaces, human perception of simulated emotions, and scoring users-confidence and certainty in the domain of problem solving in a multimodal environment.


Molecular Imaging Subgroup

Explores the static and dynamic in vivo imaging of molecules using positron emission tomography (PET) and fluorescence/bioluminescence optical tomography. We work on an integrated Bayesian framework for model-based iterative image reconstruction that combines accurate statistical and physical modeling with customized calibration and rapidly converging reconstruction methods in which we explicitly control image resolution through a single image smoothing parameter. 


Multimedia Signal Processin

Conventional content-based audio processing methods try to identify individual objects in a clip. Our research includes the investigation of ambient environmental sounds and events. Environment sounds are neither speech nor music, but comprised mainly of unstructured natural sounds that we hear everyday. Our goal is to build and design an acoustic environmental recognition system. We are interested in exploring new feature extraction, machine learning and pattern recognition algorithms related to this problem.


Multimedia Data Compression

Our recent work focuses on new techniques for multi-view video coding (MVC), with an emphasis on the joint encoding of video and depth maps for improved view synthesis, and on improved intra-prediction coding schemes. We have also developed new wavelet transforms on geometric graphs for image coding, with applications to encoding of natural images and the depth maps used in MVC. There has also been recent work in distributed video coding, among a number of other applications.


Multi-sensor and Bio Signal Processing

Recent research in this area focuses on developing new techniques for performing microarray classification. Various methods have been proposed, i.e., methods using linear discriminant analysis for analyzing microarray data, sparse representations of microarray data, etcetera. Research is also being done in Optical Coherence Tomography.



The research goal of networking group of media communication lab includes improve performance of existing wireless networks and propose innovative design for the next generation wireless systems. Current interests are peer-to-peer networking on MANET, coexistence analysis of Wi-Fi and BT and MAC solution, cognitive MAC, power-aware topology control for wireless ad-hoc networks and MAC protocol design for topology controlled wireless ad-hoc networks.



The SpeechLinks project focuses on creating robust, widely-deployable and cost-effective technology solution for enabling and enhancing cross-lingual spoken language interaction between people who do not share a common language. The state-of-the art in speech to speech translation system design is characterized by a pipelined architecture of speech recognition, machine translation and speech synthesis that relies primarily on lexical information.


SPAN (Speech Production and Articulation knowledge)

The SPAN Group is interested in using cutting-edge imaging and signal processing technologies to understand language production from its cognitive conception to its bio-mechanical execution to its signal properties. Our group uses articulator movement tracking, real-time imaging of the vocal tract, and state-of-art movement, image, and acoustic analysis techniques, including those found in automatic speech recognition paradigms.



The major interest for the video subgroup includes efficient expression of digitized multimedia data and their applications. From the generic problem of rate-distortion optimization to the advanced computer vision based video encoding, diverse techniques are covered by our researches. Other than the video encoding problems, we are also interested in the robust behaviors of networked video and application centric cross-layer optimization of video communications as well.