Embedded Signcommunication Recognition Using KNN and HMM-VITERBI Fusion Classifiers
4 Pages Posted: 1 Aug 2019
Date Written: August 1, 2019
A new hybrid approach for the Embedded Sign Language Recognition (ESLR) is proposed using the combination of the K-means NearbyNetwork (KNN) and Hidden Markov Model (HMM-VITERBI) classifiers in order to avoid the interpreter assistance. This system is combination of the hand gesture image recognition along with the Non-Audible Mumble (NAM) speech recognition system. For image recognition webcam with microcontroller unit with associated display device is used. For hand sign detection Bluetooth module is used. This type of approach will be very useful for the deaf and dump people. The proposed hybrid approach consists of two separate feature extraction techniques and one combined classification module is used to recognize the sigh language. By using this hybrid classifiers ourproposed method gives the accuracy ofabout 65% for the hand gesture images and accuracy of about 85% for the audio speech signals and the overall accuracy for both the inputs is been measured as 88%.
Keywords: Sign Language Recognition, ARM Processor, Hand Gesture, NAM microphone, KNN and HMM VITERBI classifier
Suggested Citation: Suggested Citation