Microplastic contamination in agricultural and environmental soils represents a critical challenge for ecosystem health, food security, and sustainable land management. Conventional detection methods, including microscopy and spectroscopy, are labor-intensive, destructive, and poorly suited for large-scale spatial assessment. Hyperspectral imaging (HSI) has emerged as a transformative non-destructive technology that combines high spectral resolution with spatial mapping capabilities, enabling automated detection, classification, and quantification of microplastics in complex soil matrices. This review examines the fundamental principles of HSI for soil analysis, emphasizing spectral signature characterization of common polymer types, preprocessing workflows, and advanced machine learning algorithms for automated classification. We discuss feature extraction methods, including band selection and dimensionality reduction, alongside supervised and deep learning approaches such as convolutional neural networks and support vector machines. Applications spanning laboratory validation, agricultural monitoring, and environmental risk assessment are critically evaluated. Key challenges including spectral variability due to soil heterogeneity, weathering effects, detection limits, and the need for standardized spectral libraries are addressed. Future directions emphasize integration with autonomous sensing platforms, real-time processing algorithms, and development of field-deployable HSI systems for scalable environmental surveillance and precision agriculture applications.