Abstract
Hidden Markov Model (HMM) has been used in Human Action Recognition (HAR) since early 1980s. In this paper, an improvement of HMM using fuzzy concepts is proposed and used in HAR. HMM fuzzification is used widely in some research areas such as speech recognition and medicine, but it is used in HAR rarely. To increase the classification performance and decrease information losing in vector quantization, a fuzzy approach is used in HMM implementation. In feature extraction step, a human is represented by a skeletonization method which is effective and almost real-time. Experiments show that recognition rate increases about 6 percent using this approach. We call this approach Fuzzy Hidden Markov Model (Fuzzy HMM).
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
