RGB-Hyperspectral Fusion for Macro and Microplastic Classification

The patent combines hyperspectral imaging technology with RGB image fusion technology to create a unified detection system. The hyperspectral component provides spectral information for material identification, while the RGB component provides high spatial resolution for morphological features. By merging these two imaging modalities, the system achieves both high spectral resolution and high spatial resolution simultaneously, overcoming the limitation where hyperspectral alone cannot resolve small or fibrous micro plastics.

The patent replaces traditional mechanical separation and manual identification methods with an automated optical detection system. Instead of using physical separation techniques and human operators to identify plastics, the system uses combined hyperspectral and RGB imaging with automated image processing and classification algorithms. This substitution of mechanical and manual processes with automated optical-mechanical systems dramatically increases throughput while maintaining reliable detection capability across different plastic sizes.

The patent implements a feedback-based automated classification system that uses machine learning algorithms to analyze the fused hyperspectral and RGB images. The system provides real-time feedback through automated classification, eliminating manual selection entirely. The feedback mechanism continuously refines classification accuracy by comparing detected features against known plastic signatures, achieving high classification accuracy without the time consumption and error rates associated with manual methods.