Título: Estudo para Aplicação de Redes Neurais Artificiais em um Dispositivo Detector de Posição para Medidor de Pressão Intracelular
Autores: Bertucci-Neto, Victor; Pessoa, José Dalton Cruz; Cruvinel, Paulo Estevão; Ferreira, Ednaldo J.
Resumo: In this paper is shown an application of Artificial Neural Networks for image detection. It was used in a pressure probe for vegetable cells. The pressure measurement is made by means of the detection of a meniscus in a capillary filled with silicone oil. When the tip of the capillary punctures a vegetable cell, the meniscus is displaced from its original position. By means of thermal compensation it is possible to return the meniscus back to the origin. If the power used for this compensation is known, is also known the pressure applied at the tip of the capillary. A video camera optically coupled with a microscope is used for the meniscus observation. A single video line was selected at the capillary image, showing a pulse that corresponds to the meniscus, and a logical circuit was used for the determination of the distance from the origin. However, in the presence of noise this circuit cannot sometimes recognize the meniscus pulse. Based on this, it is proposed the application of a Neural Network to recognize the meniscus pulse, even in the presence of noise. The learning rule used was the Error Backpropagation, and results are shown and discussed.
Código DOI: 10.21528/CBRN2001-124
Artigo em pdf: 5cbrn_124.pdf
Arquivo BibTex: 5cbrn_124.bib