Performance Analysis of Automatic Lip Reading Based on Inter-Frame Filtering

Abstract

Automatic lip-reading has been focused as a complimentary method of automatic speech recognition in noisy environments. One of the most competitive lip-reading algorithms is the image transform based lip-reading (ITLR) algorithm. However. ITLR has severe performance degradation under illumination variations. RASTA is a kind of inter-frame filtering method. It is used for rejecting stationary and convolutional noise in speech signal processing. In this paper., we apply RASTA approach to ITLR and analyze the performance of this method. We propose two merging techniques of pre-integration (PRE-I) and post- integration (POST-I). In PRE-I RASTA inter-frame filtering is performed ahead of the image transform process. In POST-I, inter-frame filtering is done after the image transform process.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 12, 2002
Accession Number
ADP014024

Entities

People

  • Jin-Young Kim
  • Seongmo Park
  • Seungho Choi

Organizations

  • Chonnam National University

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Automated Speech Recognition
  • Automatic
  • Bandpass Filters
  • Cognitive Science
  • Communication Channels
  • Data Displays
  • Databases
  • Engineering
  • Feature Extraction
  • Filters
  • Filtration
  • Image Processing
  • Recognition
  • Sensor Fusion
  • Signal Processing
  • Workshops

Fields of Study

  • Engineering

Readers

  • Image Processing and Computer Vision.
  • Military Engineering.

Technology Areas

  • AI & ML
  • AI & ML - Information Retrieval