| Abstract [eng] |
The topic analysed in this Master’s thesis is relevant in various modern high-tech development and application domains. As the need for smart machines and robots is increasing, scientists are trying to create the most effective method which could carry out human functions – recognizing objects, locations, and people from specific information. Modern methods of model detection are procedures based on characteristic points. Characteristic points describe the most distinctive elements of an object and place the corners and other details. In this thesis methods like SIFT, SURF, FAST with a detector SURF, SURF with a detector HOG, SURF with a detector BRISK were chosen in order to identify a location from a picture. Moreover, an overview of these methods was given as well as the description of the main characteristic points and detectors. An algorithm was developed for using MATLAB program to study the method. The performance deficiencies and inaccuracies of the experiment were identified. The logarithmic probability was calculated for location quantified estimation. |
