REAL-TIME VISUAL TRACKING SYSTEM FOR CLASSROOM SPEAKERS - PowerPoint PPT Presentation

1 / 20
About This Presentation
Title:

REAL-TIME VISUAL TRACKING SYSTEM FOR CLASSROOM SPEAKERS

Description:

Design and implement a real-time DSP system. Develop and test a dynamic and efficient algorithm ... Skin-Tone Filter. Noise Reduction. Background Mask. Cropping ... – PowerPoint PPT presentation

Number of Views:28
Avg rating:3.0/5.0
Slides: 21
Provided by: eceGa
Category:

less

Transcript and Presenter's Notes

Title: REAL-TIME VISUAL TRACKING SYSTEM FOR CLASSROOM SPEAKERS


1
REAL-TIME VISUAL TRACKING SYSTEM FOR CLASSROOM
SPEAKERS
  • Group 1
  • ECE4006
  • December 8, 2005

WaiLing Chan Vincent Lacey Lucas
Magasweran Jaimin Shah Justin Vogt
2
OVERVIEW
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Testing
  • Conclusions

3
STATEMENT OF PROBLEM
  • Design and implement a real-time speaker tracking
    algorithm

4
PROJECT GOAL
  • Design and implement a real-time DSP system
  • Develop and test a dynamic and efficient
    algorithm
  • Implement the system in real-time using a DSP
    processor and TI software

5
OVERVIEW
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Testing
  • Conclusions

6
SYSTEM OVERVIEW
  • Constraints on Real-Time implementation?

7
HARDWARE and SOFTWARE
  • Hardware
  • TI TMS320C6701 Processor
  • Sony EVI-D30 Camera
  • Software
  • Matlab - Prototyping
  • C
  • Code Composer Studio

8
TI DSP TMS320C6701 FLOATING POINT PROCESSOR
  • Launched at 167MHz 1336 MIPS and 1 GFLOPS
  • Executes up to 8 instructions in single cycle (6
    float, 2 integer)
  • 128K RAM on-chip (split 64K program / 64K data)
  • 32-bit registers
  • Efficient compilation of C code.
  • Data is byte-addressable. (It can be 8-bit,
    16-bit or 32-bit).

9
EVI-D30 CAMERA
  • High Speed, Wide Range Pan/Tilter, Conference
    Camera
  • X12 Optical Zoom, High Speed Auto-Focus Lens
  • 6 Position Preset
  • RS232C Serial Control
  • IR remote Commander
  • Time, Date Generator

10
Overview
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Testing
  • Conclusions

11
ALGORITHM
12
IMAGE PREPROCESSING
  • Bitmap to Binary Conversion
  • Skin-Tone Filter
  • Noise Reduction
  • Background Mask
  • Cropping
  • Bitmap Size 300 x 400

Original Frame Extracted
13
PROCESSING EXAMPLE
Potential Skin Tones Detected
Background Mask
14
MASKED AND CROPPED IMAGE
Cropping Image
Final Image
15
FACE DETECTION
  • Face MaskSearch for Correlation of White Pixels
    to Rectangle Representing Average Face Shape (17
    x 8)

Face Mask Zoomed to 500
Face of Speaker Zoomed to 500
16
STEADY AND PREDICTIVE MOTION
17
OVERVIEW
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Demo
  • Summary

18
OVERVIEW
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Demo
  • Summary

19
OVERVIEW
  • Introduction
  • System Overview
  • Algorithm
  • Hardware Implementation
  • Demo
  • Summary

20
QUESTIONS??
  • www.ece.gatech.edu/academic/courses/ece4006/05Fall
    /ece4006c/group01
Write a Comment
User Comments (0)
About PowerShow.com