Feasibility of Using Reflectance Spectra from Smartphone Digital Images to Predict Quality Parameters of Bananas and Papayas

Autor(es)

de Oliveira, Maisa Azarias ; Ribeiro, Michele Nayara ; Valente, Henrique Murta ; Mutz, Yhan da Silva ; Pinheiro, Ana Carla Marques ; Nunes, Cleiton Antonio

• Tipo de documento: Artigo Científico
• Fonte: FOOD ANALYTICAL METHODS; v. 17, n. 1, p. 9-pg., 2023-11-28.

Resumo

The present study aimed to build calibration models based on smartphone digital images to estimate bananas and papayas' sensory and physicochemical characteristics at different ripening stages. Three distinct image processing were evaluated: (i) the average red, green, and blue (RGB) values of the entire image, (ii) the number of pixels in each RGB value (the RGB histogram), and (iii) a reflectance spectrum mathematically obtained from the RGB triplet. Each approach was modeled with either multiple linear regression (MLR) or partial least squares (PLS) regression. The predicted variables were the sensory characteristics of ideal sweetness, ideal firmness, and global acceptance, and the physicochemical characteristics of total soluble solids (TSS) and firmness. The models obtained good responses for sensory and instrumental parameters, with the best being the modeled from the reflectance spectrum. The uniqueness of the obtained reflectance spectra captured the color changes throughout the fruit ripening. Furthermore, the obtained spectra allowed for data pre-treatments to be employed, leading to high R-2 > 0.9 and low error measured by external validation. Each modeled variable had a better performance with a specific pre-treatment. Therefore, the adoption of reflectance spectra from smartphone digital images led to a promising option to improve models' predictive performance, especially to assess fruit quality parameters. (AU)

• Processo FAPESP: 23/00474-4 - Sensores associados à internet das coisas para conexão do ambiente, genética e processamento ao perfil químico e sensorial de cafés especiais
• Beneficiário: Yhan da Silva Mutz
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado