A model of a multispectral camera that can capture images at five wavelengths-532, 612, 780, 850, and 940 nm-is designed and fabricated. The obtained multispectral images are analyzed. Algorithms of artificial neural network operation in the processing of multispectral images are discussed. A convolutional neural network for identification and classification of multispectral images in real-time is developed. (C) 2021 Optical Society of America
展开▼