Für eine korrekte Darstellung dieser Seite benötigen Sie einen XHTML-standardkonformen Browser, der die Darstellung von CSS-Dateien zulässt.


You are here

Title: Human Activity Analysis in Visual Surveillance and Healthcare

Author: Muhammad Hassan Khan

Supervisor: Prof. Dr. Marcin Grzegorzek

Reviewers: Prof. Dr. Marcin Grzegorzek  and Prof. Dr. Frank Deinzer 

Date of defence: 13.09.2018


Human activity analysis has received significant research attention in the recent years due to its applications in several fields such as monitoring, healthcare, surveillance, and entertainment. Human activities range from simple gestures e.g. hand-shaking, to the complex movements involving whole body e.g. walking, and they are usually captured with video cameras and motion sensors. The present work introduces novel techniques for human activity analysis in visual surveillance systems and also proposes a smart healthcare system to monitor the human well-being using video data. 

Biometric modalities have emerged as reliable means in recognizing the individuals using their physiological and behavioral characteristics. Gait is considered as an important biometric feature and it refers to the walking style of human. Unlike other physiological biometric modalities such as fingerprint, face, iris; gait does not require human interaction with the imaging system. In the last few years, gait recognition has received significant attention due to its promising performance in a controlled environment. However, the recent research is more focused on gait recognition in realistic environment where it is necessary to deal with the variations in gait patterns due to the change in viewing angle, carrying goods during the walk, walking-surface, clothing, etc. The first part of this work proposes few novel gait descriptors which effectively meet these challenges and can be assistive in the development of better surveillance systems. 

Movement analysis of human body parts is a fundamental step for many applications such as the detection of infantile movement disorders, analyzing the performance of athletes, and activity detection. Existing techniques are either marker-based solutions or they use wearable motion sensors to analyze the movements. The second part of this work proposes a smart healthcare system to analyze the movements of human body parts by using only the video data, without adopting any wearable sensors or markers. The aim of the proposed system is the early detection of movement disorders, and the evaluations of human's action during the therapeutic treatment. In particular, two methods are proposed to analyze the movement patterns in human body. The proposed techniques are evaluated on the benchmark datasets and the results are compared with the state-of-the-art methods to prove their effectiveness.