|
|
 |
| |
|
|
|
A Bandwidth Efficient Pareto Minimal Approach for Gesture based Video Streaming |
|
|
|
PP: 3263-3279 |
|
|
|
Author(s) |
|
|
|
S. Swarna Parvathi,
K. S. Easwarakumar,
N. Devi,
|
|
|
|
Abstract |
|
|
| Deaf, Hard-Of-Hearing and Speech-Impaired (D-HOH-SI) individuals have a specific interest in the development of
affordable high-quality videoconferencing as a means of communicating with their family members and peers using sign language.
Unlike Video Relay Service, which is intended to support communication between a caller using sign language and another party using
spoken language, videoconferencing can directly be used either between two deaf signers or between a caller using sign language and
the other using spoken language without the need of an interpreter. This paper proposes a Bandwidth Aware Gesture Based Layered
(BAGBL) framework for sign language recognition based video conferencing application. Assuming a D-HOH-SI individual at the
sender side, the proposed framework uses shape energy trajectory of hand sign gesture for video layering, aMulti-dimensionalMultiplechoice
Knapsack Problem (MMKP) based gradational hull pareto minimization heuristic called MMKP based Pareto Minimization
Heuristic for Substream Scaling (MPMHSS) and a heuristic for substream scheduling which is based on Dynamic Multilevel Priority
(DMP) called Modified DMP packet scheduling (MDMP) mechanism. At the receiving side, our framework includes an automatic
sign language recognizer to recognize the sign language gesture and a speech synthesizer to convert the recognized words to speech.
Our framework intelligently forms and selects video layers from a video sequence to maximize the video quality. Using extensive
simulation and mathematical analysis we show that the proposed solution: (i) is efficient in terms of recognition rate (ii) achieves high
radio resource utilization, (iii) maximizes the received video quality. |
|
|
|
|
|
|
|
