Analýza genetických algoritmů pro optimalizaci multiregionální soft segmentace v aplikaci na medicínská obrazová data

Abstract

This bachelor thesis deals with analysis of genetic algorithms for optimization of multiregional soft segmentations in medical image data applications. The specific purpose of segmentation method application is to analyze and model facial temperature distribution from IR image data during alcohol intoxication. The assumed intent of this analysis is a mathematical model allowing dynamic monitoring of gradual alcohol intoxication based on the temperature distribution dynamics. The theoretical part of the thesis describes methods for detection of alcohol intoxication by IR image. In addition, the activity of blood in the face after alcohol intoxication and explanation of genetic algorithms. In the practical part of the thesis, the K-means algorithm is controlled by an ABC algorithm that extracts the area of interest from the various probands. An important part of the analysis is the testing of ABC genetic algorithm parameters in order to achieve an optimized temperature distribution model.