Application of Compressive Sensing to Microwave Tomography
Abstract
Microwave tomography(MT) allows the safe, non-intrusive examination of the internal structure of an object by irradiating it with microwave signals and measuring how they get attenuated and diffracted as they propagate through the object. Measurement of these signals is difficult and slow, it depends on the object’s physical properties and the test setup used for this. MT is based on a scanning process that yields measurement data with unique characteristics that make it difficult to process using standard computational methods. As the number of measurements needed for an image with acceptable resolution increases, the complexity of processing all the data becomes overwhelming.Compressing sensing (CS) is a recent signal reconstruction technique based on numerical methods that can reconstruct an image accurately from a small number of sample measurements. This is the main motivation for applying CS to MT.However, there are several challenges to be solved in the practical application of CS toMT that have not been fully addressed, according to our literature review. These challenges are related to the non-Cartesian distribution of MT’s projection data, which does not allow the use of fast algorithms for efficient computation of transform operations; redundancy in the data produced by MT’s process, which makes it ill-conditioned system for CS algorithms; and inconsistencies in the data related to the type and sparsity of the original object image being examined. Additionally, the lack of experimental results related to the quality of image reconstruction makes it difficult to predict the performance of CS because this depends on the combination of the type of image and the measurement process of MT.This research explored the identified challenges and provides an approach to solving them. This required a detailed evaluation of the MT scanning process and characteristics of the sampling data produced by it. A procedure was developed to select from the MT measurement data a subset that can be used for image reconstruction using CS. Another procedure was developed to solve the challenge of numerical computation of transform operations needed for the CS algorithms. Computer simulations were used to verify the effectiveness of these procedures, and evaluations were made using actual measurement data from experimental MT setup, obtained from literature review. These evaluations provided an additional finding related to quality of reconstruction that could not be predicted from CS theory alone, regarding the relationship between the scanning process and data inconsistency from MT measurements. Finally, this research presents an approach to the application of CS to MT that addresses the identified challenges, and provides a framework towards future work towards achieving higher quality image reconstruction for MT.
Subject Area
Electrical engineering
Recommended Citation
Lopez, Rafael G, "Application of Compressive Sensing to Microwave Tomography" (2020). ETD Collection for University of Texas, El Paso. AAI28317109.
https://scholarworks.utep.edu/dissertations/AAI28317109