2.1 components will change accordingly. In this

2.1 Gaps Identified for the literature:

Ø  The  existing  watershed transformation results in over segmentation, sensitive to noise and high computational complexity those make it unsuitable for real time process.

Best services for writing your paper according to Trustpilot

Premium Partner
From $18.00 per page
4,8 / 5
Writers Experience
Recommended Service
From $13.90 per page
4,6 / 5
Writers Experience
From $20.00 per page
4,5 / 5
Writers Experience
* All Partners were chosen among 50+ writing services by our Customer Satisfaction Team

Ø  The existing literature doesn’t discuss about the method to improve the image texture feature.

Ø  To segment an object out of the complex environment, it is important to understand the context in which the objects find themselves. The main drawback is lack of context.

Ø  The measurement of  a color in RGB space does not represent color differences in a uniform scale. This is  yet to be analyzed.


2.2  Problem formulation/Need and significance of proposed research work:

Watershed transform has concerned with great attention in recent years as an efficient morphological image segmentation tool. The main drawback of watershed transform is over-segmentation, sensitive to noise and high computational complexity. There is high correlation among the R,G and B components. By high correlation, we mean that if intensity changes, all three components will change accordingly. In this paper after inputting the RGB image , The color channel is normalized zero to one, and this normalization process changes the pixel intensities values. Most importantly, it allows you to optimize. Testing and optimization requires traffic. Traffic gets much less useful when it’s split up into dozens or hundreds of tiny segments. As with the increase in tiny segments, it is difficult to extract the image details. To improve the marking process for watershed based segmentation we will use local Adaptive thresholding to mark the objects in efficient manner &  To compute the gradient magnitude of any image. If we segment the image by using the watershed transform directly on the gradient magnitude, it often results in “over segmentation.”. So it often requires to mark the foreground objects and background objects to obtain better segmentation. This research work improves the segmentation by integrating the improved watershed algorithm, local adaptive threshold with morphological image reconstruction and regionprops().This will help the proposed algorithm to segment the local object in efficient manner.

2.3  Objectives Defined:

1.      To implement the existing watershed transform for image segmentation

2.      To implement the improve watershed algorithm based on Adaptive thresholding and Morphological Image Reconstruction for color  image segmentation

3.      To compare the proposed modified watershed algorithm for image segmentation with existing based on following performance metrics

1.      PSNR

2.      MSE

3.      BER

4.      RFSIM









3.1 Proposed Methodology

The idea of the watershed approach for image segmentation has its origin from the natural real watershed. In the concept of this approach, the gray level image is considered as a relief with valleys and peaks. The image is firstly transformed into a morphological gradient image. Then the local “minima” in the gradient image are reduced to avoid over segmentation.


                          Regional Maxima                                           Global Maxima







Regional Minima                                                                                         Global Minima

         Fig.3.1..illustrates the concept of Maxima and Minima Functions in Image

An image can have mutiple regional maxima or minima but only a single global maxima or minima but only a single global maxima or minima. Determining image peaks and valleys can be used to create marker image that are used in morphological reconstruction.

Finding Areas of High- or Low-Intensity

The imregionalmax and imregionalmin functions identify all regional minima or maxima.
The imextendedmax and imextendedmin functions identify all regional minima or maxima that are greater than or less than a specified threshold.

The functions accept a grayscale image as input and return a binary image as output. In the output binary image, the regional minima or maxima are set to 1; all other pixels are set to 0.

Imposing a Minima

To determine specific minima (dark objects) in an image using the imimposemin function. The imimposemin function uses morphological reconstruction to eliminate all minima from an the image except the minima you specify.

To illustrate the process of imposing a minima, this code creates a simple image containing two primary regional minima and several other regional minima.

·    mask = uint8(10*ones(10,10));

·    mask(6:8,6:8) = 2;

·    mask(2:4,2:4) = 7;

·    mask(3,3) = 5;

·    mask(2,9) = 9

·    mask(3,8) = 9


Step 1. Input Image: An image passed to the proposed algorithm. Image must be either 2-D or 3-D plane i.e. grey scale or color image

Step 2. The input  image is divided into n x m sub-images blocks, where {n, m ? N| n, m > 0 and n, m<7}. Here, values for n and m are kept small because further higher for them will give the stagnant results. In minima selection process, the mask image is employed to reconstruct the image. Step 3. Compute Gradient Magnitude: Use some simple arithmetic to compute the gradient magnitude. The gradient is high at the borders of the objects and low (mostly) inside the objects. If we segment the image by using the watershed transform directly on the gradient magnitude, it often results in "over segmentation."So it often requires to mark the foreground objects and background objects  in prepared mask to obtain better segmentation.