การถอดรหัสสัญญาณมอร์สจากการกะพริบตา

Abstract

This research presents a system of eye blink Morse decoding by camera. In order to maximize both accuracy and efficiency, this system integrated many algorithms. We perform face and eye position prediction by Viola-Jones algorithm with Mixture of RGB (mRGB) color space which are used to create ROI and decrease excess searching region. Meanwhile, Kalman Filter was operating to detect eyes and shortening previous algorithm's process. Moreover, K-mean classification was fixable to assist further processes by distinguishing short and long eye blinks. These individual algorithms would function and facilitate pervious algorithms to make the most efficient way to interpret eye blink into Morse code for communication purpose. The result shows and confirms that the efficiency of this process is 94.74%. Average speed of system is 25 frames per second when works on OpenCV