Efficient Hyperspectral Data Compression using 3D Convolutional Autoencoder

Hyperspectral image has large size due to abundance of spectral bands preventing efficient storage and transmission in real-time applications. This paper introduces a 3dimensional convolutional autoencoder (CAE) for hyperspectral data compression, utilizing CAE for both compression and subsequent de...

Full description

Saved in:
Bibliographic Details
Published in:2024 3rd International Conference on Advancement in Electrical and Electronic Engineering (ICAEEE) pp. 1 - 5
Main Authors: Afrin, Afsana, Mamun, Md. Al
Format: Conference Proceeding
Language:English
Published: IEEE 25.04.2024
Subjects:
Accuracy
Convolution
Deep Learning
HSI Compression
Image coding
PSNR
Recurrent neural networks
Solid modeling
Three-dimensional displays
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Hyperspectral image has large size due to abundance of spectral bands preventing efficient storage and transmission in real-time applications. This paper introduces a 3dimensional convolutional autoencoder (CAE) for hyperspectral data compression, utilizing CAE for both compression and subsequent decompression. Our method achieves improved Compression Ratio (CR) values of 982 and 960 on the KSC and Botswana datasets, along with better Peak Signal-to-Noise Ratio (PSNR) results of 55.29 and 52.64, surpassing existing 2D convolutional autoencoder-based HSI compression. Our approach also shows negligible differences in classification accuracy between original image and the reconstructed version of the compressed image, confirming its effectiveness in real-world applications.
DOI:10.1109/ICAEEE62219.2024.10561855