Hand Gesture Recognition system for physically challenged people using IoT

One of the rising fields of research are hand gesture recognition. Being an important part of non-verbal interaction is a critical part of our everyday life. Hand management recognition systems allow us to communicate with the machine, which is more familiar to the human being, creative, normal, user-friendly. Hand gesture recognition covers a wide range of applications including contact with the human machine, sign language, immersive game technology etc. To perform this we have implemented a method of hand recognition and extraction of features by means of a web camera in real time. The picture is captured in this method through the system's webcam. The initial processing of the input image and thresholding are used to remove image noise and smooth the image. Following this, apply region filling in the gesture or object of interest to fill holes. The “HU moments” are used for feature extraction and classification using KNN algorithm. This helps in improving the classification though machine learning and recognition step. We have tested 43 gestures which include 26 gestures of English alphabets and achieved a confidence of 99.9%.