Title :
Method of deformable optimum threshold for lipreading
Author :
Hongxun, Yao ; Rui, Wang ; Wen, Gao
Author_Institution :
Dept. of Comput. Sci. & Technol., Harbin Inst. of Technol., China
Abstract :
This paper presents a new approach to the location and/or extraction of mouths during lipreading preprocessing. In other words, a new approach to making binary lip images with a deformable optimum threshold is presented. Using this approach makes the mouth object follow more closely the true shape of the mouth, and can improve the features from the image data. It is useful for improving the rate of recognition. The deformable optimum threshold technique and its application is introduced in this paper
Keywords :
feature extraction; gesture recognition; optimisation; speech recognition; automatic speech recognition; binary lip images; binary segmentation; convolution; deformable optimum threshold; image data; lipreading; lipreading preprocessor; mouth object; Automatic speech recognition; Computer science; Convolution; Data mining; Face recognition; Image processing; Image segmentation; Loudspeakers; Mouth; Shape;
Conference_Titel :
Signal Processing Proceedings, 1998. ICSP '98. 1998 Fourth International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-4325-5
DOI :
10.1109/ICOSP.1998.770760
