|
|
Image similarity measurement using points of interest
Jelínek, Ondřej ; Uher, Václav (referee) ; Burget, Radim (advisor)
This paper presents a new object detection method. The method is based on keypoints analysis and their parameters. Computed parameters are used for building a decision model using machine learning methods. The model is able to detect object in the picture based on input data and compares its similarity to the chosen example. The new method is described in detail, its accuracy is evaluated and this accuracy is compared to other existing detectors. The new method’s detection ability is by more than 40% better than detection ability of detectors like SURF. In order to understand the object detection this paper describes the process step by step including popular algorithms designed for specific roles in each step.
|
|
|
Profiling a State? Combining the Methods of Micro-Level Profiling with the Scope of Macro-Level Strategic Analysis
von Werdt, Benjamin Max David ; Peacock, Timothy (advisor) ; Rosa, Paolo (referee)
This dissertation presents a new method of strategic analysis - the Profiling Cycle. It proposes to combine micro-level profiling methods as used in criminal profiling, behavioral psychology, and sociology, with macro-level methods used for strategic analysis, including neoclassical realism and a newly developed comparative strategy. The Theory of the Big 5 was chosen to represent behavioral psychology, and Coleman's Boat was used for sociology. Key elements from these prevailing theories and methods were borrowed and applied on the level of an entire state. The dissertation explains the Profiling Cycle step-by-step, thereby allowing anyone wishing to use this new method to do so. For illustration, it applies the first step of the Profiling Cycle to the Russian Federation as a case study. The Profiling Cycle's ability to accurately predict future behavior can only be tested in the medium to long term. This dissertation acts as a proof-of-concept. It describes how a functionally complete profile can be created with the help of behavioral psychology, and how an analyst might use this profile to set up possible medium to long term scenarios. Keywords: Strategic Analysis, Theory of the Big 5, Coleman's Boat, Criminal Profiling, Russia, New method, Profiling Cycle
|
|
|
ESTIMATION OF EQUILIBRIUM HYDROGEN PRESSURE - A NEW METHOD
Čermák, Jiří ; Král, Lubomír
A new method is proposed to estimation of hydrogen pressure in equilibrium with hydride phase in a hydrogen\nstorage material. It is applicable both for hydrogen absorption and desorption in cases where the hydride phase\nis formed by nucleation and growth mechanism. The proposed method saves considerably the experimental\ntime replacing the conventional time consuming measurement of pressure-composition isotherms, the so\ncalled PCT curves. The proposed evaluation procedure is illustrated using hydrogen chemi-sorption at\ntemperatures 623 K, 573 K and 523 K in chosen hydrogen storage alloys Mg-Si-C, Mg-Li-C and Mg-Na-C.
|
|
|
Estimation of equilibrium hydrogen pressure - A new method
Čermák, Jiří ; Král, Lubomír
A new method is proposed to estimation of hydrogen pressure in equilibrium with hydride phase in a hydrogen\nstorage material. It is applicable both for hydrogen absorption and desorption in cases where the hydride phase\nis formed by nucleation and growth mechanism. The proposed method saves considerably the experimental\ntime replacing the conventional time consuming measurement of pressure-composition isotherms, the so\ncalled PCT curves. The proposed evaluation procedure is illustrated using hydrogen chemi-sorption at\ntemperatures 623 K, 573 K and 523 K in chosen hydrogen storage alloys Mg-Si-C, Mg-Li-C and Mg-Na-C.
|
|
| |
|
|
Image similarity measurement using points of interest
Jelínek, Ondřej ; Uher, Václav (referee) ; Burget, Radim (advisor)
This paper presents a new object detection method. The method is based on keypoints analysis and their parameters. Computed parameters are used for building a decision model using machine learning methods. The model is able to detect object in the picture based on input data and compares its similarity to the chosen example. The new method is described in detail, its accuracy is evaluated and this accuracy is compared to other existing detectors. The new method’s detection ability is by more than 40% better than detection ability of detectors like SURF. In order to understand the object detection this paper describes the process step by step including popular algorithms designed for specific roles in each step.
|
guest ::
