Capturing micro-vibration images in plants caused by homeopathic application

• Autor(a) principal: Zanco, Jasper Jose
• Data de Publicação: 2022
• Outros Autores: Boff, Pedro, Domingues, Sérgio, Boff, Mari Ines Carissimi
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Agropecuária Catarinense (Online)
• Texto Completo: https://publicacoes.epagri.sc.gov.br/rac/article/view/1226
• topic: agrohomeopathy; vibraimage; electric tension; computational vision; agrohomeopatia; vibraimage; tensão elétrica; visão computacional

The use of images, sensors and mathematical algorithms can help in the generation of technical attributes and facilitate the plant health diagnosis. Combined with this, computer vision provides a non-destructive and non-invasive strategy for collecting samples and analyzing plant propagules, provided the experiment traceability. Thus, the objective of this research was to identify signs of homeopathies of Magnetitum and Arsenicum tartaricum applied in purslane [Pilea microphylla (L.) Liebm.], using computational algorithms. The work of images capturing was carried out in the Laboratory of Plant Production and Didactic Garden of the Agronomy Course, UNISUL University. To evaluate signs in plants, based on the images, algorithms found in VibaHT® and ImageJ were used. The images were generated by webcam (online) and two homeopathies at 250 milesimal were applied for 14 days. The experimental procedure consisted of generating “640 × 480” pixel images from a transformed webcam to simulate a "red-green-NIR" (RGN) sensor, replacing the channel with a blue light filter and thus produce a near-infrared image (NIR). The images were also generated in their normal "red-green-blue" (RGB) channels to test the algorithms' competence. After capturing the images, mathematical analyzes of the pixel’s variation were performed, represented by three variables, developed by specific algorithms: lacunarity, entropy and stress. The number of experimental repetitions was sufficient to identify significant differences at the 1% probability level between the images, and the algorithms were robust to identify the signs of homeopathy.