Single image non-uniformity correction using compressive sensing |
| |
Affiliation: | 1. Computer Science, Vanderbilt University, Nashville, TN, United States of America;2. Electrical Engineering, Vanderbilt University, Nashville, TN, United States of America;3. Biomedical Engineering, Vanderbilt University, Nashville, TN, United States of America;4. Psychiatry and Behavioral Sciences, Vanderbilt University Medical Center, Nashville, TN, United States of America;5. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, TN, United States of America;6. Laboratory of Behavioral Neuroscience, National Institute on Aging, MD, United States of America |
| |
Abstract: | A non-uniformity correction (NUC) method for an infrared focal plane array imaging system was proposed. The algorithm, based on compressive sensing (CS) of single image, overcame the disadvantages of “ghost artifacts” and bulk calculating costs in traditional NUC algorithms. A point-sampling matrix was designed to validate the measurements of CS on the time domain. The measurements were corrected using the midway infrared equalization algorithm, and the missing pixels were solved with the regularized orthogonal matching pursuit algorithm. Experimental results showed that the proposed method can reconstruct the entire image with only 25% pixels. A small difference was found between the correction results using 100% pixels and the reconstruction results using 40% pixels. Evaluation of the proposed method on the basis of the root-mean-square error, peak signal-to-noise ratio, and roughness index (ρ) proved the method to be robust and highly applicable. |
| |
Keywords: | Single image non-uniformity correction Compressive sensing Sampling/reconstruction |
本文献已被 ScienceDirect 等数据库收录! |
|