Publication: Implementation of integer karhunen– loève transform on raspberry pi 4 and openjpeg for lossless hyperspectral compression
Loading...
Date
2021-07-01
Authors
Vinayagan, Sarveswaran
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The research work presented in this paper is concerned with the implementation of the Karhunen-Loève Transform (KLT) in Raspberry Pi 4 board for hyperspectral image compression. The Integer KLT is used for spectral decorrelation due to its superior performance in decorrelating the spectral components in hyperspectral images compared to other transforms used for image compression and OpenJPEG is used as the spatial compressor. The OpenJPEG is only applicable to desktop platforms therefore spatial compression could not be done on the Raspberry Pi 4 and the OpenJPEG will be evaluated on the desktop platform. The objective of this research is to implement an Integer KLT algorithm on the Raspberry Pi 4 board and evaluate its performance by comparing it to other embedded platforms as well as evaluating the performance of the OpenJPEG as spatial compressor. The performance of the algorithm on the board in terms of execution time is evaluated. The execution times are compared to other embedded platforms from previous works. Clustering technique is used to improve the execution time as well reduce the computational complexity of the image. The execution time for the compression of the AVIRIS and Hyperion hyperspectral image datasets improve with increasing number of clusters. The Raspberry Pi 4 has shown to have greater processing capabilities as compared to the Beaglebone-Black and the multi core DSP TMDSEVM6678L.