Enhanced image-based algorithm for construction of three dimensional zapin animation from monocular video

• Main Author: Azmi, Nik Mohammad Wafiy
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: QA76 Computer software
• Online Access: http://eprints.utm.my/102974/1/NikMohammadWafiyMSC2022.pdf.pdf; http://eprints.utm.my/102974/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:150717
• Institution: Universiti Teknologi Malaysia

Extracting human motion data from body posture or movements has gained popularity in computer and multimedia industry nowadays. Industries such as entertainment, gaming and Artificial Intelligence demand the technology that can provide them with the most fast and accurate motion capture technology available. This is due to the high needs of the motion data which can be used to increase the naturalness of the graphic contents and also the automation process involving human interaction. However, the technology of the motion capture is very expensive due to its requirement for proper space and high-end equipment. In addition, the motion capture technology available today require substantial amount of time in order to complete one production of the motion data this is not entirely because of the limitation of the devices and system but mostly because the technology needs the help of specific experts to be operational. Therefore, this research addressed the issue by introducing an improved algorithm in image-based process to extract the motion data as an alternative to the motion capture technology. The approach of the research is to use an enhanced image-based algorithm to improve previous image-based processing technique in generating 3D parameters which can be used to create 3D animation. The research was also done with the intention of proposing a new methodology in preserving the Intangible Cultural Heritage in Malaysia. The general framework of the research includes three main phases namely features/keypoints extraction, visualization of 3D skeleton and lastly evaluation of research’s results. Outcome from these processes is a 3D skeleton along with motion information extracted from monocular video. The evaluation of the research consists of qualitative and quantitative analysis which involved the videos and results comparison, keypoints evaluation and expert testing. All these evaluations were made in order to testify the accuracy and the satisfaction towards the results of the research. The research managed to produce an accurate 3D skeleton animation based on the movements of the Zapin dance.