Desenvolvimento de um equipamento para captura de imagens multiespectrais

Abstract

The development of multispectral imaging systems has become increasingly relevant for applications that require rapid identification and classification of materials, particularly in scenarios where conventional solutions are associated with high cost and limited portability. In this context, this work presents the development of a lowcost and portable system for multispectral image acquisition based on a Raspberry Pi platform. The main objective is to provide a compact solution capable of capturing spectral information at different wavelengths, enabling preliminary in situ analysis of a variety of samples. The adopted methodology involves the use of a light-emitting diode panel with eight wavelengths distributed across the ultraviolet and nearinfrared regions, operating in a multiplexed manner to illuminate the scene. Image acquisition is performed using an RGB-IR microcamera controlled by a Raspberry Pi board, on which a basic image processing algorithm was implemented to enable visualization of reflectance data and spectral histograms for each band. The developed system was experimentally evaluated in different application scenarios, including the identification of damaged or overripe fruits, the classification of different types of plastic materials, and the analysis of water-related properties. The obtained results demonstrate that the proposed prototype is capable of providing relevant spectral information for classification tasks, showing satisfactory performance considering its cost and architectural limitations. It is concluded that the proposed solution represents a viable and accessible alternative when compared to commercial multispectral imaging systems, standing out for its portability, flexibility, and potential applicability in different scientific and industrial fields.