Sie sind auf Seite 1von 8

MID TERM REPORT OF

COLOR IMAGE SEGMENTATION

A thesis to be submitted

to

MANIPAL UNIVERSITY
(Formerly known as Manipal Academy of Higher Education)

For Partial Fulfillment of the Requirement for the


Award of the Degree

of

BACHELOR OF ENGINEERING

in

ELECTRONICS AND COMMUNICATION ENGINEERING


by
RICHA TOMAR
Reg. No. 070907082
Final Year B.E.

Under the Guidance of

Mr. T.G.S.CHANDRASEKHARAPPA Ms.K.Poornima


Professor, Department of E &C Engineering Assistant professor
Internal Guide External Guide / Co Guide
DEPARTMENT OF ELECTRONICS AND
COMMUNICATION ENGINEERING
MANIPAL INSTITUTE OF TECHNOLOGY
(A constituent college of Manipal University, Manipal)
MANIPAL - 576 104, KARNATAKA, INDIA

MARCH 2011
COLOR IMAGE SEGMENTATION

Abstract — Image segmentation is a classic subject in the field of image processing and also is a hotspot and focus of image
processing techniques. With the improvement of computer processing capabilities and the increased application of color image,
the color image segmentation are more and more concerned by researchers. Color image segmentation methods can be seen as
an extension of the gray image segmentation method in color images, but many of the original gray image segmentation methods
can not be directly applied to color images. This project proposes a color image segmentation method of automatic seed region
growing on basis of the region with the combination of the watershed algorithm with seed region growing algorithm.

Keywords— Electroglottography (EGG), Artificial Neural Network (ANN), Back propagation algorithm

1. INTRODUCTION

People are only interested in certain parts of the image in research and application of the image. These parts are
frequently referred to as a target, they generally correspond to the image in a specific and unique nature of are. It
needs to extract and separate them in order to identify and analyze object, on this basis, it will be possible to further
use it for the target. Image segmentation is a technique and process which divides the image into different regions
and extract out the interested target. Here features can be pixel grayscale, color, texture etc. Predefined targets can
correspond to a single or multiple regions. To illustrate the level of image segmentation in image processing, we
have introduced “image engineering” concept, it brings the involved theory, methods, algorithms, tools, equipment
of image segmentation into an overall framework.

1.1 Image engineering is divided into 3 levels:


1.1.1 Image Processing
1.1.2 Image Analysis
1.1.3 Image Understanding

1.1.1 Image Processing


It is the emphasis on the transformation between images and improves the visual effects of images.
It is relatively low-level operations.
It is mainly operated on pixel level.

1.1.2 Image Analysis


It is mainly monitoring and measuring the interested targets in the image in order to get it’s objective
information and as a result, build up a description of the image.
It is middle level operations.
It focuses on measurement, expression and description of target.

1.1.3 Image Understanding


It is the further study of the nature of each target and the linkage of each other.
It is mainly high level operations.
It focuses on the operation and illation of data symbol which abstracts from the description.

Image segmentation is a key step from image processing to image analysis. As the image segments, target
expression based on segmentation allows feature extraction and parameter measurement that converts original image
to more abstract and compact form. Thus, it is possible to make high level image analysis and understanding.
2. METHODOLOGY

There are many methods for implementing color image segmentation, for instance: histogram threshold, region-
based approach, based on edge detection methods, fuzzy methods, artificial neural network approach etc.

In this project, we use the method of region growing along with watershed transform.

The basic idea of region growing is a collection of pixels with similar properties to form a region. First, we need to
find a seed pixel as a starting point for each needed segmentation and then merge the same or similar property pixels
with the seed pixel around the seed pixel domain. These new pixels act as new seed pixels and this above process
continues until no more pixels satisfy the conditions specified for region growth.

Even though the method of region growing is easy to complete and compute, it is rarely used alone. It is usually
combined with other segmentation methods to overcome the disadvantages of region growing which are as follows:

 It needs human interaction to obtain seed point(user needs to implant a seed point in every region it needs
to extract).

 The pattern of regional growth is also sensitive to noise.

This project specifies certain rules to automatically select seed pixels, thus, solving the first problem. Also carrying
on regional growing on the basis of watershed algorithm solves the problem of noise.

The overall procedure for the entire project has been customized as:

Conversion of color space for image

Using watershed method to initialise segmentation

Automatically selected seed region

Region growing

Consolidation of region
So far,we have reached till the third step in the procedure i.e. automatic selected seed region. Watershed algorithm
has been successfully implemented to help in automatic seed selection.
1. Conversion of color space for image

A color model is an abstract mathematical model describing the way colors can be represented as tuples of numbers,
typically as three or four values or color components (e.g. RGB and CMYK are color models). However, a color
model with no associated mapping function to an absolute color space is a more or less arbitrary color system with
no connection to any globally-understood system of color interpretation.

Adding a certain mapping function between the color model and a certain reference color space results in a definite
"footprint" within the reference color space. This "footprint" is known as a gamut, and, in combination with the
color model, defines a new color space. For example, Adobe RGB and sRGB are two different absolute color
spaces, both based on the RGB model.

Color space conversion is the translation of the representation of a color from one basis to another. This typically
occurs in the context of converting an image that is represented in one color space to another color space, the goal
being to make the translated image look as similar as possible to the original.
2. Using Watershed Method to initialize segmentation

Watershed: A grey-level image may be seen as a topographic relief, where the grey level of a pixel is
interpreted as its altitude in the relief. A drop of water falling on a topographic relief flows along a path to
finally reach a local minimum. Intuitively, the watershed of a relief correspond to the limits of the adjacent
catchment basins of the drops of water.

Cardiac MRI Image Gradient Image

Relief of the gradient Watershed of the gradient

There are many different algorithms to compute watersheds. The watershed transform applies these ideas to gray-
scale image processing in a way that can be used to solve a variety of image segmentation problems.
First algorithm to compute watershed: Distance Transform

The distance transform of an image is a relatively simple concept: It is the distance from every pixel to the nearest
non-zero valued pixel. The distance transform can be computed using IPT function bwdist.
Second method: Using Gradients

The gradient magnitude is often used to preprocess a gray-scale image prior to using the watershed transform for
segmentation. The gradient magnitude image has high pixel values along object edges, and low pixel values
everywhere else. Ideally, then, the watershed transform would result in watershed ridge lines along object edges.
Third Method: Marker-controlled Watershed Segmentation

Direct application of the watershed transform to a gradient image usually leads to oversegmentation due to noise and
other local irregularities of the gradient. The resulting problem can be serious enough to render the result virtually
useless. A practical solution to this problem is to limit the number of allowable regions by incorporating a pre-
processing stage designed to bring additional knowledge into the segmentation procedure.

An approach used to control oversegmentation is based on the concept of markers. A marker is a connected
component belonging to an image. We would like to have a set of internal markers, which are inside each of the
objects of interest, as well as a set of external markers, which are contained within the background. These markers
are then used to modify the gradient image using this process. Various methods have been used for computing
internal and external markers, many of which involve the linear filtering, nonlinear filtering, and morphological
processing.
Which method we choose for a particular application is highly dependent on the specific nature of the images
associated with that application.

Das könnte Ihnen auch gefallen