Stabilizace videa na Raspberry Pi

Abstract

This bachelor’s thesis focuses on the design and implementation of a digital video sta- bilization algorithm intended for real-time processing on the Raspberry Pi 4B platform. Digital stabilization is essential for improving visual quality and enabling reliable image data processing in environments with limited computational resources. The work outlines the basic principles of digital stabilization and selected computer vision techniques that form the core of the implementation. The focus was placed on simplicity and computational efficiency. The implementation was tested on Raspberry Pi 4B. The results show that the pro- posed approach is capable of delivering stabilized video with low distortion and sufficient smoothness for real-time performance, even under constrained hardware conditions.

This thesis focuses on the analysis and implementation of digital video stabilization methods, specifically aimed at real-time processing on the Raspberry Pi 4B platform. Digital stabilization is crucial for achieving higher visual quality and effective analysis of image data, especially in environments with limited computational resources. The work emphasizes the principles and classification of digital stabilization algorithms and includes an overview of computer vision techniques utilized in this field. Based on the conducted tests and analyses, an algorithm was selected that ensures a stable image with minimal geometric distortion while being sufficiently efficient for operation on a device with limited computational power. The results of this work have practical significance for applications requiring real-time digital video stabilization on embedded systems.