Umm Al-Qura University

Collage of Computer and Info. Systems

Computer Engineering Department

Automatic Camera Tracking System

IMPLEMINTATION

CONCLUSION This project is for improving the camera systems

that capturing many speakers talk using audio signal

processing techniques to determine the active speaker

and rotate the camera toward him. It can be used in

many fields (Classrooms, online conference, meeting

room). In the future we will integrate the system into

one device with the same performance which will

make it portable and more efficient.

ABSTRACT

The project idea is to design a system that

can be able to capture video stream for more

than one speaker using only one auto steering

camera controlled by a computer program. This

project can be used in many fields like online

meetings ,conferences and classrooms. This

idea implemented based on the Sound Source

localization techniques (SSL) which use the

sound’s amplitude to identify the current

speaker. The project aim to improve the camera

recording systems in cases that contain several

speakers. The main tasks are first select the

speaker and then directing the camera towards

him during the capture period and repeat the

process when another speaker start talking.

MOTIVATION

Traditional camera systems are low efficiency

and high cost in the case of multiple speakers,

because they need a camera to each speaker and

director for the coordination between the

cameras. As computer engineers we aim to

facilitate and improve the people’s life by

providing solutions using advanced technology

to solve problems. This gives us the motivation

to design a computer system that overcome this

problem and offers new capabilities for human-

computer interaction.

• Mohammed Yaseen Almanna

• Saleh Hadi Al-Shahrani

• Ali Al-Hatmi

• Abdulrahman Mahjoub

• Mohammed Al-Malki

SUPERVISORDr. Majid Al-Gethami

The core processing power take place in a MATLAB

program. Matlab use Digital Signal Processing (DSP) and

Data Acquisition (DAQ) Toolboxes to process the audio

stream signals. By cutting the received signals into FRAMES

then apply quantization techniques we can find out the

highest peak in each frame that allow us to determine the

highest amplitude using DSP methods, after applying these

steps frame by frame in loops called steps, the algorithm find

out the speaker.

After many experiments and tests we reach the highest

accuracy decision that made our project accurate, simple and

efficient with error rate less than 1%.

The program divided into 2 Phases:

First Identify the microphones and assign the parameters that

considered as the keys for an accurate decision making.

Identify the arduino and the servo motor and assign the

control pins.

Second apply the (Step) Process on the three audio streams.

TEAM & SUPERVISOR

1- 3 microphones capture the audio and send the signals to the computer using USB Hub.

2- The USB Hub receive the signals and send it to PC in serial connection.

3- PC receive signals from USB Hub and apply the processing algorithms on them using MATLAB

determine the speaker by calculating the highest amplitude then send the speaker position to Arduino

microcontroller.

4- When Arduino receive the speaker ID signal it rotate the servo motor toward the speaker direction using

predefined angels based on number of speakers which represent the microphones constant directions.

5- A camera fixed on servo motor face the speaker and send the video to the PC using serial connection.

SYSTEM DESIGN

Figure1 : Stream Framing Process

Figure2 : First Audio Signal

Figure3 : Second Audio Signal

Figure4 : Third Audio Signal

BLOCK DIAGRAM

COMPARISON