Arun S. Narayanan, Lam W. Cherk, Tan K. Kiong, Koh C. Huat, and Clarence W. de Silva


Tele-health, video streaming, adaptive streaming, selective-streaming


In this paper, the design of a video streaming system for telehealth applications is presented. Deviating from general-purpose objectives of off-the-shelf teleconferencing solutions, the system is customized to enable doctors and patients to be virtually linked over a broad range of connection speeds yet maintaining a satisfactory video/audio streaming experience to carry out remote diagnosis and assessment. Specifically, a key objective is to maintain a frame rate of around 15 fps at various connection speeds through an appropriate selection and characterization of transport protocols, encoding technologies and other innovations. Through the use of the Microsoft’s Kinect sensor, under low bandwidth conditions, the system allows a dynamic cropping of human body structure disregarding all other background information. Images are much smaller in size but sufficient for certain applications like telrehabilitation. The system also allows selective cropping for the user to discard the background but send only the useful information. Finally, through the use of adaptive streaming approach, the system is able to adapt the image quality to the network speed while maintaining the frame rate. Collectively, with these customized innovations incorporated, the system is able to perform video calls at frame rate of around 15 fps even in 128 kbps network.

Important Links:

Go Back