Kinect Gait Based AuthenticationCategories
Course & option:
Diploma in Infocomm Security Management
Kinect Gait Based Authentication
(left to right according to the photo)
Muhammad Sholihin Bin Kamarudin, Kenneth Kan, Bai Qing Rong, Zahir Ahmad
Dr Lu Li Ming (LU_LIMING@SP.EDU.SG)
Kinect, Kinect SDK 1.8, Processing2.0 ( Programming)
The ability to produce an automated detection system to identify users based on their unique gaits using a Kinect is one of a good security implementation the world can have. Gait patterns are always made unique due to the fact that different subjects always having a different walking behaviour. Research from the past have proven and shown that Gaits are hard to hide and imitate. One of the greatest advantage enables Gait recognition to not require the subject’s attention during the implementation phase as the system would run behind the scenes and detect any possible unauthorized users, alerting the management for further action. These are potential advantages that should be heavily considered.
The project aspires to identify subjects without intrusion into the subject’s daily life. Gait is an important part of every human life. An algorithm is set in place to calculate length of joints together with an angle to form an identification method. It aims to research if an algorithm can be calculated based on the diverse findings. This solution is designed to work as a background process. Research has shown that it is indeed possible to make a security program with a Microsoft Kinect. The Kinect has been equipped with Depth cameras which will enhance the results of the data to make the program more accurate.
In attempt to evaluate if the program is suitable to be used as a security implementation, various testing and experimentations will be carried out. This is targeted at economical organisations, such as schools, where a Kinect may prove useful and cost effective. Most importantly, this program is targeted and proposed for its ability to be non-intrusive. Through experimentations, proper documentation are anticipated to achieve results. Profiling is a critical phase in this project. It helps to gather distinctive data to be used for research.
The team made use of Microsoft Kinect sensor to research and explore more possibilities. The Microsoft Kinect has already been programmed to be able to detect and calculate XYZ. The proposed areas include just the left portion of the human body. Data will be extracted. In this program, the team will be using Processing, a language that is similar to Java. Data for comparisons will be stored in MySQL database.
So imagine, when a person were to enter a class, they would have to walk pass the Kinect at a normal walking speed. The Kinect will do the following, for each millisecond after a person is detected, it will record the X&Y Axis of the persons limbs. For each millisecond it will continue to record and once the user has pass the Kinect fully is where the magic happens. The Kinect will do many types of calculation such as calculating average speed of the user walk, length of limbs, unique gait movements, and angles between the arm during hand swing movement and also save a recording of the user walk.
All this will then be compared against the database and ensuring that the matching has a 92% similarity rate from the obtained calculation to the stored calculation which is obtain during the profiling phase.
This project has a success rate of 87% of identifying the user.
This program can be used in replacement of the Attendance taking system. Precautionary warning letters have been sent from the school to inform students that sharing of attendance code is strictly prohibited and yet these actions still persist. With the widespread of social media and communication platforms which the school has no jurisdiction against; should they decide to take disciplinary measures something which most students are aware of.
With our program in place and gait being unique to each individual, it can’t be shared among students thereby tackling the issue: sharing of codes.