Dimensionality Reduction of Hyperspectral Imagery

ISBN-13: 9783031426667 / Angielski / Twarda / 2023

Arati Paul;Nabendu Chaki

Opis

This book provides information about different types of dimensionality reduction (DR) methods and their effectiveness in hyperspectral data processing. The authors first explain how hyperspectral imagery (HSI) plays an important role in remote sensing due to its high spectral resolution that enables better identification of different materials on the earth’s surface. The authors go on to describe potential challenges due to HSI being acquired in hundreds of narrow and contiguous bands, represented as a 3-dimensional image cube, often causing the bands to contain information redundancy. They then show how processing a large number of bands adds challenges in terms of computation complexity that reduces efficiency. The authors then present how DR is an essential step in hyperspectral data analysis to solve these issues. Overall, the book helps readers understand the DR processes and its impact in effective HSI analysis.

Spis treści

Introduction.- Remote sensing.- Digital image processing.- Hyperspectral image characteristics.- Dimensionality reduction.- Dataset description.- Pooling based band extraction.- Ranking based band selection.- Band optimization.- Data Driven approach.- Conclusion.

O autorze

Arati Paul is a Scientist in Regional Remote Sensing Centre - East, National Remote Sensing Centre, Indian Space Research Organisation. She has completed B.Tech, followed by M.Tech in computer science and Engineering. She has received her Ph.D. from University of Calcutta. Her area of work includes remote sensing, GIS, image processing and geospatial data analytics. She has more than 60 publications including research papers, book chapters and technical reports in her area of expertise. She is also involved in outreach activities of ISRO and delivered talks on several occasions in different workshops/ conferences and training programmes of the centre. 

Nabendu Chaki is a Professor in the Department Computer Science & Engineering, University of Calcutta, Kolkata, India.  He is sharing 7 international patents including 4 US patents. Besides editing more than 30 conference proceedings with Springer, Dr. Chaki has authored 7 text and research books and over 250 Scopus Indexed research papers in Journals and International conferences. He has served as a Visiting Professor in different places including Naval Postgraduate School, USA; Ca Foscari University, Italy and AGH University in Poland. He is the founder Chair of ACM Professional Chapter in Kolkata and served in that capacity for 3 years since January 2014. He was active during 2009-2015 towards developing several international standards in Software Engineering and Service Science as a Global (GD) member for ISO-IEC.

Blurb

This book provides information about different types of dimensionality reduction (DR) methods and their effectiveness in hyperspectral data processing. The authors first explain how hyperspectral imagery (HSI) plays an important role in remote sensing due to its high spectral resolution that enables better identification of different materials on the earth’s surface. The authors go on to describe potential challenges due to HSI being acquired in hundreds of narrow and contiguous bands, represented as a 3-dimensional image cube, often causing the bands to contain information redundancy. They then show how processing a large number of bands adds challenges in terms of computation complexity that reduces efficiency. The authors then present how DR is an essential step in hyperspectral data analysis to solve these issues. Overall, the book helps readers understand the DR processes and its impact in effective HSI analysis.

• Presents a data driven approach for dimensionality reduction (DR);
• Discusses the effect of spatial dimension and noise in the context of DR of hyperspectral imagery (HSI);
• Includes an optimization based approach for DR challenges and identification of gap areas in existing algorithms along with suitable solutions.