Hyperspectral Image Classification: Potentials, Challenges, and Future Directions.

Area of Science:

• Remote Sensing
• Image Processing
• Artificial Intelligence

Background:

• Hyperspectral imagery and remote sensing are advancing rapidly.
• Intelligent technologies like machine learning (ML) and deep learning (DL) enhance processing of complex, high-resolution images.
• These methods offer precision and fidelity for analyzing multi-band, 3D data.

Purpose of the Study:

• To systematically review machine-dependent technologies and deep learning for hyperspectral landcover classification.
• To identify research gaps and formulate key questions through a novel review approach.
• To provide an organized overview and assessment of contemporary ML methods in hyperspectral image analysis.

Main Methods:

• Systematic literature review of WoS, Scopus, SCI, and SCIE indexed articles.
• Focus on machine learning (ML) techniques including support vector machines, sparse representations, active learning, extreme learning machines, transfer learning, and deep learning.
• Analysis of advancements in ML for hyperspectral image identification.

Main Results:

• Identification of key machine learning and deep learning contributions to hyperspectral landcover classification.
• A structured overview of current ML methods applied to hyperspectral data.
• Assessment of the literature highlighting the capabilities of these technologies.

Conclusions:

• Machine learning and deep learning significantly enhance hyperspectral image analysis for landcover classification.
• The systematic review provides a foundation for understanding current research and future directions.
• Researchers can leverage this work to deepen their knowledge of ML applications in hyperspectral remote sensing.