Lip-Contourbased Speaker Activity Detection in Smart Environment


Downloads per month over past year

Raheja, JL and Sharma, S and Plinge, A and Fink, GA (2013) Lip-Contourbased Speaker Activity Detection in Smart Environment. In: 14 th IEEE International Vacuum Electronics Conference (IVEC-2013), May 21-23, 2013, Paris. (Submitted)

PDF - Submitted Version
Download (371Kb) | Preview


In this paper we describe a lucid model for efficient localization of an active speaker in a real-time video for a video conference. The system is built on a simple and robust method of contour detection of grey images and frame differencing of binary images for motion detection. This proposed method is independent of luminosity and removes unnecessary parameters from the frame to be worked upon so that the processing time is reduced and a faster algorithm is obtained to detect the speaking person. We aim at developing a robust technique for talker identification by processing a continuous video input from a live web camera and removing the necessity of using the cumbersome process of identifying a speaker in a discussion or a video-conference and then focusing on the person manually. Results of testing the suggested model are presented with conclusions

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Speaker, Lips, Contour Detection, Frame Differencing
Subjects: Electronic Systems > Digital Systems
Divisions: Electronic Systems
Depositing User: Mr. Rabin Chatterjee
Date Deposited: 22 May 2013 04:45
Last Modified: 22 May 2013 04:45

Actions (login required)

View Item View Item