Skip to main content
ARS Home » Southeast Area » Athens, Georgia » U.S. National Poultry Research Center » Quality and Safety Assessment Research Unit » Research » Publications at this Location » Publication #410236

Research Project: Assessment of Quality Attributes of Poultry Products, Grain, Seed, Nuts, and Feed

Location: Quality and Safety Assessment Research Unit

Title: ML-assisted multispectral and hyperspectral imaging

item Yoon, Seung-Chul
item EKRAMIRAD, NADER - Oak Ridge Institute For Science And Education (ORISE)

Submitted to: Book Chapter
Publication Type: Book / Chapter
Publication Acceptance Date: 11/3/2023
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Hyperspectral imaging (HSI) holds great significance due to its ability to provide rich spectral and spatial information, enabling a comprehensive understanding of the composition and characteristics of materials or scenes. Unlike conventional imaging methods such as color imaging, HSI generates an abundance of spectral data across a continuous range of wavelengths, creating a 3D data cube. These hyperspectral images, often composed of hundreds of spectral wavelengths, are captured using specialized devices like pushbroom line scanning cameras with spectrographs or cameras equipped with liquid crystal tunable filters. However, before task-specific machine learning (ML) models can be applied, these images necessitate further processing. ML techniques form the core of hyperspectral image processing and analysis, as they extract valuable features and information embedded within these images. During the preprocessing stage, hyperspectral data undergoes transformation and processing for analysis using ML methods. The subsequent processing and analysis stages heavily depend on ML models. The final phase of HSI involves postprocessing. Among the multitude of applications, we particularly highlight the utilization of ML and artificial intelligence (AI) techniques for evaluating food quality and safety through hyperspectral image analysis. Additionally, we explore the strengths and challenges associated with employing these learning approaches to analyze the intricate hyperspectral data for food inspection.