Advances in feature selection methods for hyperspectral image processing in food industry applications: a review

Area of Science:

• Food science
• Spectroscopy
• Imaging technology

Background:

• Hyperspectral imaging (HSI) offers rich spatial and spectral data for food analysis.
• The high dimensionality of HSI data presents computational challenges, known as the curse of dimensionality.
• Feature selection is crucial for efficient and accurate analysis of HSI food data.

Purpose of the Study:

• To review feature selection algorithms for hyperspectral imaging in food applications.
• To categorize and explain different search strategies used in feature selection.
• To guide future research in HSI data processing for food quality, safety, and authenticity.

Main Methods:

• Categorization of feature selection algorithms into complete, heuristic, and random search.
• Review of fundamental principles of each algorithm type.
• Illustration of algorithm applications in food hyperspectral data analysis.

Main Results:

• Discussion of the advantages and disadvantages of various feature selection methods.
• Identification of key algorithms applicable to hyperspectral food data.
• Highlighting the impact of feature selection on computational load and predictive accuracy.

Conclusions:

• Feature selection is vital for optimizing hyperspectral imaging applications in the food industry.
• Understanding different search strategies aids in algorithm selection.
• This review provides a framework for effective feature selection in food HSI analysis.