Digitální zpracování a analýza obrazu teplotního pole s využitím Fourierovy transformace

Abstract

In the industrial practice, where construction and validation of prototypes of new products research is being conducted, plays an increasingly important role methods based on numerical modeling. This trend can be understood primarily with regard to funding, which must be made for the production and testing of the prototype in real terms, or conditions close to reality. Much cheaper and more time efficient is development of mathematical model of the part and its subjugation to mathematical simulations. Essential for achieving results, corresponding with acceptable accuracy with reality, is an entire knowledge of all initial and boundary conditions of the simulation. This "adjustment" of model can’t be made explicit in most cases. To achieve conformity with reality, it’s necessary to gradually refine the model parameters with measurement. This thesis deals with the verification of parallel physical - mathematical modeling and focuses on the evaluation of the results using the methods of digital image processing. The dissertation are at first summarized and evaluated general trends of the main low-level digital image processing algorithms applied to image of the temperature field, with an emphasis is placed on approaches using Fourier transform. Performance of algorithms is monitored by established objective metric. Subsequently, there are introduced the numerical experiments and methods designed for evaluation of obtained results. In conclusion are summarized results of the experiments and also the direction of possible further development of this work.