Tomographic microwave diversity image reconstruction employing unitary compression

• Published in: IEEE transactions on microwave theory and techniques Vol. 40; no. 2; pp. 315 - 322
• Main Authors: Zhao, Z., Farhat, N.H.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01-02-1992; Institute of Electrical and Electronics Engineers
• Subjects: Applied sciences; Data compression; Data mining; Exact sciences and technology; Feature extraction; Frequency; Image coding; Image processing; Image reconstruction; Information, signal and communications theory; Karhunen-Loeve transforms; Microwave imaging; Microwave theory and techniques; Signal processing; Telecommunications and information theory; Tomography; Covariance; Theoretical study; Microwave; Tomography; Algorithm; Information compression; Image reconstruction; Mean square error
• ISSN: 0018-9480; 1557-9670
• DOI: 10.1109/22.120104

Data compression through a unitary transform is utilized in tomographic microwave diversity image reconstruction in order to reduce the dimensionality and to extract the features in the data space. The unitary compression is derived by minimizing the mean-square error (MSE) and the unitary transform is made of eigenvectors of the data's covariance, regarded as a Karhunen-Loeve transform. Tomographic microwave frequency-swept imaging was developed using a unique target-derived reference technique to access the three dimensional Fourier space of the scatterer and an image reconstruction algorithm based on the projection slice theorem. It is shown that centimeter resolution of a complex object can be preserved even when half of the data set is compressed and that the reconstructed image remains identifiable by a human observer even when 2/3 of the data set is compressed.< >