Projects

• Syntactic/Semantic String Matching - This project involves the development of invariant curve matching schemes and their applications to the silhouettes of marine mammals for the identification of an individual in large databases of photos.
  
  
• Digital Embedded Watermarking - Watermarking is the process by which we hide an image in a document or other image in order to track and establish copyright information. This project is trying to establish some bounds on the number of bits that can be used and the tradeoff with compression capability.
  
  
• Optimized and Robust Wireless Communications - The successful evolution of the future high-speed wireless communication networks require significant advances in adaptive signal processing algorithms. This research involves the development of a general and novel framework for designing robust and spectrally efficient channel estimation, synchronization and equalization algorithms.
  
  
• Iteratively Decodable Codes for Wireless Communications and Magnetic Recording - This project is currently investigation design and analysis of irregular turbo codes and other graph based codes such as low density parity check codes for applications in next-generation wireless communications applications and high density magnetic recording. The objective is to design low complexity coding schemes which are well matched to the iterative (turbo) decoding algorithm.
  
  
• Monte Carlo Signal Processing for Wireless Communications - This project investigates the design methodologies of adaptive optimal receivers in wireless channels. The approach is to formulate the problems of signal reception in unknown time-varying channels as multivariate Bayseian interference problem. Monte Carlo signal processing methods are employed to develop adaptive systems for computing Bayesian estimates of the channel and data. An array of receiver design problems found in wireless communications are being treated under this framework.
  
  
• Applications of Stereo Imagery to Differential Geometric Measures of Robustness - This project combines stereo imaging with theoretical mathematics (viz. differential geometry) for application within the areas of communications and signal processing.
  
  
• Denoising and Compression of Large Remote Sensing SAR Images - There is a need to remove multiplicative speckle noise in a large synthetic aperture radar (SAR) image before compressing and downloading to a ground station for remote sensing studies. This project uses the wavelet frame/SOT structure coupling with anisotropic diffusion/Markov Random Fields to efficiently accomplish this goal. The algorithm can be carried out with a fast DSP engine at near real time speed in a satellite in space.
  
  
• Modulation and Coding Techniques for Space-Time Systems - A promising technique to increase spectral efficiency as well as improve performance in a wireless communication system is the use of multiple transmitter/receiver antennae. This project considers the design and analysis of improved modulation and coding techniques for these so-called space-time systems. Starting from fundamental principles of communication theory, new design criteria are being developed which result in improved system design.
  
  
• 3-D Embedded Subband Coding with Optimized Truncation (3-D ESCOT) - This project develops an efficient video coding algorithm: 3-D embedded subband coding with optimized truncation (3-D ESCOT), in which coefficients in different subbands are independently coded using fractional bit-plane coding and candidate truncation points are formed at the end of each fractional bit-plane. A rate-distortion optimized truncation scheme is used to multiplex all subband bitstreams together into a layered one. A novel motion threading technique is also used to form threads along the motion trajectories in a scene. For efficient coding of motion threads, memory-constrained temporal wavelet transforms are applied along entire motion threads. Block-based motion threading is currently implemented in conjunction with 3-D ESCOT in a real video coder. Extension of 3-D ESCOT to object-based coding also is addressed. Experiments demonstrate that 3-D ESCOT outperforms MPEG-4 for most test sequences at the same bit rate.
  
  
• Analysis and Modeling of Gene Expression Data from cDNA Microarrays - Microarrays provide simultaneous measurement of the expression levels of thousands of genes. Expression levels are measured by an image processing algorithm applied to a digital image of the microarray. The measurements are used to construct classifiers to discriminate among cancers (and other disease types) at the molecular level. They also are used to build models of genomic networks that characterize the communication system within the genome.