Nowadays, Kinect is a well-known skeleton and motion detecting technology that
uses in many fields of work, e.g., computer games, interactive works in exhibitions or
museums. Unfortunately, one Kinect can detect in an area around 2 square meters . For a large
-scale installation, multiple modules of Kinects are needed.
According to our survey, there are many companies or groups of people already
implemented APIs, or frameworks for multi-Kinect detection, but it is proprietary and not yet
readily available as a commercial product. Our project focuses on the study of an overlap area
between two Kinects, using a Kalmon’s filter to calculate an efficient area of overlapped. An
experiment of set up factors (such as light) is also provided and summarized into a manual to
help developers to set up their projects. For the detection part, blob detection is a tool
coordinated with depth detection. Moreover, a communication between client and server is
implemented using socket programming.
In the end, we develop a program that provide a framework for client-server
architecture for multiple Kinects modules. Users can create a server by running our provided
application for server on Windows system. For client, an implementation is embeded into
Kinect depth detection that provided by Microsoft. Users need to run our program to create
client. And for the set-up multiple modules, we provide a manual that illustrated in Appendix
A. Users can follow our instructions to set up the modules.