Browsing by Author "Chikkanna, M."

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Kinect based real-time gesture spotting using HCRF
    (2013) Chikkanna, M.; Ram Mohana Reddy, Guddeti
    The sign language is an effective way of communication for deaf and dumb people. This paper proposes, developing the gesture spotting algorithm for Indian Sign Language that acquires sensory information from Microsoft Kinect Sensor. Our framework consists of three main stages: hand tracking, feature extraction and classification. In the first stage, hand tracking is carried out using frames of Kinect. In second stage, the features of Cartesian system (velocity, angle, location) and hand with respect to body are extracted. K-means is used for extracting the codewords of features for HCRF. In the third stage, Hidden Conditional Random Field is used for classification. The experimental results show that HCRF algorithm gives 95.20% recognition rate for the test data. In real-time, the recognition rate achieves 93.20% recognition rate. � 2013 IEEE.
  • No Thumbnail Available
    Item
    Kinect based real-time gesture spotting using HCRF
    (2013) Chikkanna, M.; Guddeti, G.
    The sign language is an effective way of communication for deaf and dumb people. This paper proposes, developing the gesture spotting algorithm for Indian Sign Language that acquires sensory information from Microsoft Kinect Sensor. Our framework consists of three main stages: hand tracking, feature extraction and classification. In the first stage, hand tracking is carried out using frames of Kinect. In second stage, the features of Cartesian system (velocity, angle, location) and hand with respect to body are extracted. K-means is used for extracting the codewords of features for HCRF. In the third stage, Hidden Conditional Random Field is used for classification. The experimental results show that HCRF algorithm gives 95.20% recognition rate for the test data. In real-time, the recognition rate achieves 93.20% recognition rate. © 2013 IEEE.