Vous êtes sur la page 1sur 27

1

IMPLEMENTATION OF LIVE VIDEO CARTOONING


USING THRESHOLDING AND MULTICHANNEL
MOTION DETECTION ON DSP PROCESSORS
PROJECT GUIDE: PROJECT MEMBERS: Batch no:40
Dr. M. PREMA KUMAR M.Tech., Ph.D T. N. V. V. S. LAKSHMI (15B01A04B9)
Associate Professor, R. ANANTHA LAKSHMI (15B01A04B6)
Dept. of ECE. V. SRILAKSHMI (15B01A04H2)
Y. G. S. L. SOWMYA (15B01A04H7)

Department of Electronics and Communication Engineering

SHRI VISHNU ENGINEERING COLLEGE FOR WOMEN : BHIMAVARAM


(AUTONOMOUS)
2
OUTLINE

 Motivation  Multi Channel Motion Detection


 Objectives  Motion Detection Process Thread
 Introduction  Implementation of Motion Detection
 Literature Survey  Results for Thresholding
 Working for Cartooning  Results for Motion Detection
 Block Diagram for Thresholding  References
 Flow Chart for Thresholding  Conclusion
3
MOTIVATION

 In our day to day life in order to spread information and communicate with each
other we use image as one of the medium.

 Sharing videos has become very popular. Not only do people use videos to
simply share their day-to-day lives with others, video sharing has become a
powerful tool for exposing fraud and improving accountability of public officials.
 In cluttered environment motion detection and object tracking plays a major role.
 To do image processing we choose DSP processor.
4
OBJECTIVES

 To create a code for implementing cartoonish features in a real time


video on a live video.

 To implement “Multichannel Motion Detection on DSP Processor”.


5
INTRODUCTION

 Thresholding is a simple, yet effective way of partitioning an image


into foreground and background. In the project we apply thresholding
on a video in YUV format which is captured from CCD camera.
 Motion detection is the process of detecting a change in position of an
object relative to its surroundings or the change in the surroundings
relative to an object. In the project we observe motion detection by
determining the difference between the frames.
6
LITERARURE SURVEY

 Sylvain Paris, Pierre Kornprobst et al “Bilateral Filtering Theory and


Applications”Fredo Durand. Foundations and TrendsR in Computer Graphics
and Vision[1].
This paper deals with blurring of an image while respecting strong edges. The
limitation of the bilateral filtering is that it can’t be performed on live video.
 Xiao Chen“Application of Matlab in Moving Object Detecting Algorithm. 2008
International Seminar on Future BioMedical Information Engineering[2].
This paper deals with video processing steps such as frame separation, binary
operation, gray enhancement, filter operation, detection and extraction of
moving object on images according to frame difference-based dynamic-
background refreshing algorithm.
7
WORKING FOR CARTOONING

 Here we use single thresholding, which means only one threshold


limit is specified.
 So, first plan on the thresholding limits as to convert the video.
 The input video can be split into three parts; they are the intensity
component ‘Y’ and the two chrominance components ‘U’ and ‘V’.
 The desired cartoonish effect can be obtained on the video just by
varying the intensity in each frame.
8
CONTD…

 The cartooned video which is in YUV format is converted into


RGB format by using following equations
R = Y + 1.402 (V-128)
G = Y - 0.34414 (U-128) - 0.71414 (V-128)
B = Y + 1.772 (U-128)
9
BLOCK DIAGRAM FOR CARTOONING

Video
Video Capturing Daughter
Scene Device (CCD Card Display
Camera)

TMS320C6416 Processor

YUV signals taken


Threshold on video YUV signal
from Video
YUV signals converted to RGB
Daughter Card
FLOW CHART FOR THRESHOLDING

Start

Video captured from camera

Initialize Pixel Coordinates

Initialize pixel coordinates

Compare pixel value with threshold

No
If pixel>128

Yes

Make pixels 245

Make pixels 10

Stop
11
MULTICHANNEL MOTION DETECTION

 Motion detection is done by comparing the current video frame to a


specified reference frame. Using a pixel-by-pixel comparison,
unequal portions of the current frame are colored.
 Reference Framework 5 (RF5) for eXpressDSP Software provides
a fast ramp to DSP application development using DSP/BIOS and
the eXpressDSP Algorithm Standard (XDAIS).
12
MOTION DETECTION PROCESS THREAD
13
IMPLEMENTATION OF MOTION DETECTION

 ADDING DIFF ALGORITHM, INTEGRATING CELLS, APPROPRIATE


SOURCE AND HEADER FILE
 We can describe the operation of the algorithm very easily:
1. Compare all pixels in a certain block and count the number of
differences
2. If there are more than DIFFTHRESHOLD (user-defined) differences,
go to 3. Else, stop.
3. Find all different pixels and change the color value.
14
CONTD…

 PROCESSING CIF FRAMES


 Camera captures a CIF YUV stream
15
CONTD…

 De-multiplexing into Even and Odd Y, U, and V Buffers


16
CONTD…

 Prescale from 4:2:2 to 4:2:0


17
CONTD…

 CONV_YUY2toIYUV
18
CONTD…

 CREATING A PROCESSING CHANNEL


1. Add the appropriate enumerations to thrProcess.h
2. Modify the ThrProcess structure in thrProcess.h
3. Open each channel
4. Execute the channel
 ADDING CELL TO A PROCESSING CHANNEL
 Cells give us a convenient and consistent way to wrap an algorithm and insert it
into RF5. The Channel Manager (CHAN) then uses the wrapper as an interface
for efficient execution.
19
CONTD

 RESETTING THE REFERENCE FRAMES AND CHANGING DIFF COLORS


20
RESULTS FOR THRESHOLDING

Input video
21
CONTD…

Cartooned Video
22
CONTD…

When threshold limit =150 When threshold limit =110


23
RESULTS FOR MOTION DETECTION

Motion Detection Output


24
CONTD…

Motion Detection Output


25
REFERENCES

[1] Sylvain Paris, Pierre Kornprobst proposed et al “Bilateral Filtering Theory and
Applications”, Foundations and Trends in Computer Graphics and Vision, 2008.
[2] Xiao Chen “Application of Matlab in Moving Object Detecting Algorithm”
International Seminar on Future BioMedical Information Engineering, 2008.
[3] K. Martin Sagayam, Nitin John James et al “Implementation of Live Video
Cartooning Using Thresholding Technique in YUV Format” International Journal of
Advanced Research Trends in Engineering and Technology (IJARTET) Vol. 1, Issue
1, September 2014.
[4] A DSP/BIOS AIC23 Codec Device Driver for the TMS320C6713 DSK,
(SPRA677), June 2003.
26
CONCLUSION

 The project “Implementation of Live Video Cartooning Using


Thresholding and Multichannel Motion Detection on DSP
Processors” is successfully implemented and tested. We performed
motion detection using DIFF algorithm by observing the change in
the pixel values of the current and previous frames of the live video
and the detected output is displayed on LCD display and we
obtained the cartoon like appearance of the live video feed by using
thresholding technique with less computational overhead.
27

Vous aimerez peut-être aussi