FACE RECOGNITION AND LIP READING FOR PASSWORD AUTHENTICATION

In recent years several speech recognition systems that use visual together with audio information showed a significant increase in performance over the standard speech recognition systems. The use of visual features is justified by both the bi-modality of the speech generation and by the need for features that are invariant to acoustic noise perturbation. The audio-visual speech recognition system design using Matlab can be used for password authentication. This project designed in two modules i)Face Recognition Module and ii)Lip Reading module.

In the first module, Input images are trained to find the Face Recognition. In this module, a picture of the speaker is captured and it is matched to the images of 10 different speakers in the database. If the matching is successful, "SPEAKER IDENTIFIED" message is displayed and the control is transferred to the next module. Otherwise, "INVALID SPEAKER" message is displayed and the system terminates.

The next module is Lip Reading module. For each of the 10 speakers present in the database, a particular word is assigned. For example - Speaker 1 - JUSTICE Speaker 2 - BUCOLIC and so on...

In this second module, a video of the recognized speaker is taken as input. The ROI (here, the lips) is extracted and the movements are analyzed using PIXEL based method (DWT) and Hidden Markov Models. For example, if the recognized speaker from the first module is SPEAKER 2, then the lip movements are compared with the Model for the word BUCOLIC. Suppose there is a match, authorization is successful and "SPEAKER IDENTIFIED" message is displayed. Otherwise, if there is no match, "AUTHORIZATION FAILED" message is displayed. For each of the 10 speakers present in the database, a particular word is assigned. For example - Speaker 1 - JUSTICE, Speaker 2 - BUCOLIC and so on...

A video of the recognized speaker is taken as input. The ROI (here, the lips) is extracted and the movements are analyzed using PIXEL based method (DWT) and Hidden Markov Models. For example, if the recognized speaker from the first module is SPEAKER 2, then the lip movements are compared with the Model for the word BUCOLIC. Suppose there is a match, authorization is successful and "SPEAKER IDENTIFIED" message is displayed. Otherwise, if there is no match, "AUTHORIZATION FAILED or NOT IMAGE" message is displayed.

Database Words Used in Design:

  • Word

  • Justice

  • Serendipity

  • Pleasure

  • Pride

  • Bucolic

  • Trumpet

  • Intellect

  • Fleece

  • Champion

Reference Paper-1: A Coupled HMM for Audio-Visual Speech Recognition

Author’s Name: Ara V. Nefian, Luhong Liang, Xiaobo Pi, Liu Xiaoxiang, Crusoe Mao and Kevin Murphy

Source: Microcomputer Research Labs, Intel Corporation

Year:Unknown

Reference Paper-2: Automated Lip Reading Technique for Password Authentication

Author’s Name: Sharmila Sengupta,Arpita Bhattacharya,Pranita Desai and Aarti Gupta Mao and Kevin Murphy

Source: International Journal of Applied Information Systems (IJAIS)

Year:2012

Request source code for academic purpose, fill REQUEST FORM or contact +91 7904568456 by WhatsApp or info@verilogcourseteam.com, fee applicable.

SIMULATION VIDEO DEMO