How to calculate Threshold Lip Reading using Matlab

7 views (last 30 days)
Hi,
I am new for matlab , as per my requirement i have implement lip key points like left ,right,upper ,bottom and centroid .with reference link http://www.mathworks.in/matlabcentral/answers/54610 after that how calculate threshold for lip reading .
example : If I Silent Spell word Hello with the help of boundary lip key points how to track threshold match known template?
Best example let me know .Thanks
T_Sum_Pixel = sum(bw_videoFrame1);
bw_threshold = 350000;
if T_Sum_Pixel > bw_threshold
h = msgbox('Operation Completed','success');
end

Accepted Answer

Image Analyst
Image Analyst on 28 Dec 2013
I don't know the answer, and probably no one here does. The people who know the answer can be found here: in VisionBib
21.3.4 Lipreading, Lip Reading, Lip Tracking
21.3.4.1 Combined Audio Visual Recognition and Analysis
21.3.4.2 Mouth Location, Lip Location, Detection
  2 Comments
Image Analyst
Image Analyst on 30 Dec 2013
Edited: Walter Roberson on 6 Jan 2014
SAMEER's "Answer" moved here because it's not an answer, but a reply to me:
Thanks for reply my question , i have go through various pdf documents , but it's difficult to understand how they are implement. because i am newbie for matlab . example with reference link 1: Lip Reading another lip reading reference link 2: Matlab Lip Reading I think link 2 is easy way to achieve please let me know any idea about lip reading.
Thanks
Image Analyst
Image Analyst on 30 Dec 2013
Sorry, I just don't have time to get the paper, read and understand it, code it up, and explain it to you. I don't even have time to try all the things I'm interested in, much less for anyone else. Good luck though. You'll probably learn more doing it yourself anyway.

Sign in to comment.

More Answers (0)

Categories

Find more on Introduction to Installation and Licensing in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!