ABC algorithm based optimization of 1-D hidden Markov model for hand gesture recognition applications

Hand gestures are extensively used to communicate based on non-verbal interaction with computers. This mode of communication is made possible by implementing machine learning algorithms for pattern recognition. A stochastic mathematical approach is used to interpret the hand gesture pattern for classification. In this work, a predominant method is used by 1-D hidden Markov model (1-D HMM) for classifying the patterns and to measure its performance. During training phase, 1-D HMM is used to predict its next state sequence of hand gestures using dynamic programming methods such as Baum-Welch algorithm and Viterbi algorithm. However, dynamic programming based prediction methodologies are complex. To enhance the performance of 1-D HMM model, its parameter and observation state sequence must be optimized using bio-inspired heuristic approaches. In this work, Artificial Bee Colony (ABC) algorithm is used for optimization. A hybrid 1-D HMM model with ABC optimization has been proposed which has yielded a better performance metrics like recognition rate and error rate for Cambridge hand gesture dataset.