Super-Resolution Target Detection and Tracking.
Abstract
A computational efficient blind deconvolution algorithm has been developed which recovers an information bearing signal that has been distorted by transmission through an unknown system. In the array processing problem, a target detection and location algorithm which provides quality estimates in the presence of impulsive type noise has been developed. Its performance significantly improves upon existing algorithms. The blind deconvolution algorithm is based on a kurtosis analysis of the measurement data. The innovative aspect of this analysis results in one having to solve a fixed point problem. A computation efficient algorithm for solving this fixed point problem has been developed. Numerical experimentation has shown that the proposed blind deconvolution algorithm provides for a more effective deconvolution operation in comparison to existing techniques. In many target detection and location problems, the array's sensor signals are corrupted by impulsive-type noise which causes most existing direction-of-arrival algorithms to either fail or to provide unacceptably poor performance. To overcome this, a modification of the author's signal subspace DOA algorithm has been made. This algorithm is useful for general array geometries and is applicable to applications in which the incident sources are incoherent, coherent, or a mixture of incoherent and coherent sources.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1993
- Accession Number
- ADA307013
Entities
People
- James A. Cadzow
Organizations
- Vanderbilt University