Recognition of hand gestures of English alphabets using HOG method

Purpose of this paper is to develop a hand gesture recognition system which can identify hand gestures of alphabets with resemblance. Feature extraction methods play an important role in gaining high recognition rate for a gesture recognition system. Generally, there could be difficulties in recognizing gestures with resemblances and differentiating these gestures with resemblances indicate the superiority of an extraction method. This paper attempts to evaluate Histograms of Oriented Gradients (HOG), the feature extraction method for its potential to identify gestures with resemblance. KNN classifier was used for gesture recognition. System performance is evaluated using multiple statistical measures. As by the evaluation, HOG can identify gestures with an accuracy of 96%. The single handed static sign language alphabets are taken for recognition purpose.