Hand Gesture Recognition Using Color Markers

In the last few years, many methods have been proposed for gesture recognition but even now, there is not a standard method set for this type of detection. Both online and offline approaches can be used for this purpose. This paper uses the offline method with the aim to detect the 26 letters of the American Sign Language (ASL). First of all, a glove with 3 color markers, namely red, green and yellow is worn upon the hand. Then the training and testing sets are acquired. Then, masking of the three colors is done followed by noise removal. The next step is to find centroids of the white regions in each mask. Afterwards, distances between centroids are calculated and then an algorithm is developed based on the number of regions in red, green and yellow masks and the distances between centroids to detect the letters. This method has an overall accuracy of 96%.