Abstract: Human beings interact with a lot of things around them. One such interaction happens between humans and computers. As the technology and software are becoming complex day-by-day, the modes of interaction with computers are also varying. It may be through speech to text conversation, using sensors like accelerometers and gyros worn on the body of the interactor, using facial gestures, computer interfaces like keyboard and mouse, or by using external device like smart-phones or using hand gestures. Gesture based recognition system involves tracking a person’s movements and predicting what they may indicate. For every gesture recognized, an application program is launched which the user can use it to interact with the OS. The programs launched may be command prompt, notepad, and MS paint or file explorer. This hand gesture recognition algorithm uses AForge.Net framework for capturing frames from the webcam in real time. The frames captured are then processed for detection of the hand. The features are extracted and the gestures are identified to interact with the OS and launch different applications. The gestures include patterns like “V”, “M”, STOP and Triangle. The application program is designed in C# with .Net Framework 4 and it uses threading support to implement a parallel version of the algorithm. For the program to be compiled the compiler is set to allow unsafe code to run which is needed to accelerate the image processing steps. A frame is captured per second, the region is identified, the features are computed and hand gesture is identified and if no gestures are found, a message will be displayed. The hand gesture recognition rate using the above algorithm is 88.16%.

Keywords: Binarization, Threshold Filtering, Feature Extraction, Framework, Parallel Version, Compactness, Protrusion Ratio.