A yuv image edge detection method based on histogram equalisation

BY

ABUBAKAR SADIQ MUHAMMAD

MEVLANA UNIVERSITY KONYA,TURKEY

Email: alsiddiq_uhd@yahoo.com

Introduction

Colour space & models

RGB image edge detection

YUV image edge detection

Analysis of Results

Conclusions

INTRODUCTION Edge detection is an

important aspect of image processing whose goal is extracting edges in an image.

It is often the first step in image segmentation

A lot of researches

have proposed many arithmetic on RGB(colour) image edge detection

However a number of shortcomings were

noted that include

-low speed of processing of each image

-colour losses after processing of each

image

This paper proposes a method of image

edge detection using YUV colour space

and histogram equalisation

The RGB colour space is the simplest colour space that comprises the primary colours Red, Green and Blue

This colour space explores a wide range of colour

when mixed together among which are

YELLOW(RG)

MAGENTA(RB)

CYAN(GB)

The drawback of this model is that it does not suit the intuition(insight) of human psychology

Human intuition bring with it the aspect of lightness of the colour and the amount we use to colour a specific region

Why?

Also the distance

between colour points is

not equal to the vision

characteristics

Thus, it is not easy to obtain the Hue, Saturation and

Brightness attribute in an RGB image

YUV model The YUV model defines a

colour space in terms of one luma and two chrominance (U,V) components. The basic characteristics of these model is that each component is independent of the other

RGB

Y

U

V

Luminance(lightness) : is a measurement of the eye’s

perception of light intensity (brightness).

Luma: is the component of a digital image that carries

a monochrome portion that determines image

lightness- it is often defined as gamma corrected

luminance

Chrominance: stands for the colour components

obtained by deducting the luminance value Y from R

and B

The fact that human visual systems is more sensitive to

difference in lightness than in colour makes the model

application in video standard.

Application

- Used in PAL, SECAM and composite colour video

standards

The colour difference U and V in YUV

colour space are given by the equation

U= 0.493(B -Y)

V=0.877( R –Y)

So that the conversion between RGB and

YUV is as given by the equation

0.299 0.587 0.114

0.14713 0.28886 0.436

0.615 0.51499 0.10001

Y R

U G

V B

R

G

B

1 0.000 1.1400

1 -0.369 -0.581

1 2.029 0.000

Y

U

V

=

The edge detection method employs the

use of three filters

Gradient filter

Laplacian filter

Laplacian with control parameter(α)

each RGB component is processed

independently using the respective filter

Analysing the RGB image Each R,G and B component is computed Each component is separately processed using

the horizontal and vertical sobel operator Each component is separately processed using

the Laplacian operator and Laplacian with control parameter(α)

The resulting RGB image is obtained from the

separately processed components for each operator

The resulting RGB image obtained from

the separately processed components for

each operator is then histogram

equalised.

The basic idea is to process the Y component in the YUV image

The YUV is obtained from the RGB image Each component Y, U & V are computed

from the YUV image The Y component is processed using the

respective filters listed earlier The Y component is then histogram

equalised and the corresponding RGB is obtained

Obtain YUV image from

corresponding RGB

Compute each component

of YUV image

Process the Y component

using respective filters

Obtain the corresponding

RGB image of the HISTEQ

YUV image

Obtain the histogram

equalisation of the Y

component

Flow chart representation of YUV image processing

ORIGINAL RGB IMAGE & YUV IMAGE

Horizontal & Vertical sobel on each RGB component

H & Vertical sobel on Y component of YUV image

Observe that single component in YUV produces equivalent of 3 components in RGB

Laplacian / Laplacian(α) in RGB

Laplacian / Laplacian(α) in YUV

Observe that a single component in YUV produces effect of 3 RGB component

Resulting processed images in RGB

Resulting processed Y components images in YUV

Observe that in RGB images losses its colour when compared to YUV after processing

Histogram EQU on RGB images

Histogram EQU on YUV images

Observe that the colour edges are still visible in YUV as compared to RGB with HISTEQ applied

Corresponding HISTEQ YUV in RGB

It can be observed that the detected

edges are more exact based on the

proposed algorithm

The processing is faster and simple using a single component(Y-YUV) as compared to 3 –RGB components

The colour edges are also detected

effectively when compared to the former

Thanks for

listening