Reference Contour Recognition Technology on Image Using Neural Network and Neuro-fuzzy Modeling

A technology for selecting and recognizing a reference contour in an image is proposed, which consists of two stages. The first stage is based on the development of a neuro-fuzzy model of the process of selecting contour points and a neural network classifier. The input of the neural network classifier is the contour features calculated using the Fourier transform, which are invariant to displacement and scaling. An algorithm for constructing a training sample is proposed, which makes it possible to classify the contour, regardless of its size and position in the image. The second stage is associated with the use of a neuro-fuzzy model of the process of selecting contour points and recognition of a reference contour using a neural network classifier—perceptron.