The Method of Neuro-fuzzy Calibration of Geometrically Distorted Images of Digital X-Ray Tomographs

Abstract

AbstractReviewed classification of digital radiography systems for the detection of X-rays, the main quality parameters of radiographic images Calibration is the process of segmentation of distorted image in a finite number of clusters in which each pixel of the image associated with a certain degree of conditioning with the crystal, which was struck by gamma ray. The physical foundations and interpretation of the neuron learning procedure in Kohonen fuzzy cellular neural networks have been investigated. On their basis, a method for automatic calibration of digital X-ray radiography systems is described, which is reduced to neuro-fuzzy segmentation of a distorted image into a finite number of clusters and proposed a neural network-based technique for the automatic calibration of digital X-ray tomographs that do not require user intervention. To improve the calibration accuracy, Kohonen’s adaptive fuzzy clustering networks are used. The architecture of Kohonen’s three-layer fuzzy cellular neural network (CNN-SOM) is proposed.KeywordsCalibrationNeural networkAl NetworksFlat-Panel DetectorKohonen’s fuzzy cellular neural network