Colour Based Image Segmentation Using L*A*B* Colour Space Based On Genetic Algorithm

International Journal of Emerging Technology and Advanced Engineering

156

Colour Based Image Segmentation Using L*A*B* Colour

Space Based On Genetic Algorithm

Mr. Vivek Singh Rathore1, Mr. Messala Sudhir Kumar

, Mr. Ashwini Verma

1_{M.Tech Scholar 4}th _Sem., 3_{Associate Professor, LNCT, Indore(M.P.)} 2_{Assosiate Professor, CEC Bilaspur (C.G.)}

Abstract—In colour based image segmentation is made to overcome the problems encountered while segmenting an object in a complex scene background by using the colour of the image. After pre-processing, the image is transformed from the RGB colour space to L*a*b* space. Then, the three channels of L*a*b* colour space are separated and a single channel is selected depending upon the colour under consideration. Next, genetic based colour segmentation is performed on the single channel image after which practical is applied to the image to obtain the particular object of interest. As can be seen from the expected results shown in this paper the proposed method is effective in segmenting the complex background images, these results are used to propose a new colour image segmentation method. The proposed method searches for the principal colours, defined as the intersections of the homogeneous blocks of the given image. As such, rather than using the noisy individual pixels, which may contain many outliers, the proposed method uses the linear representation of homogeneous blocks of the image. The paper includes comprehensive mathematical discussion of the proposed method and expected results to show the efficiency of the proposed algorithm.

Keywords—Image segmentation, RGB colour space,

L*A*B* colour space, Separate channels, Genetic algorithm.

I. INTRODUCTION

Image segmentation may be defined as a technique, which partitions a given image into a finite number of non-overlapping regions with respect to some characteristics, such as gray value distribution, texture distribution, etc. The objective of dividing an image into homogeneous regions remains a challenge, especially when the image is made up of complex textures. Traditional methods for image segmentation have approached the problem either from localisation in class space using region information, or from localisation in position, using edge or boundary information [1]-[5]. Some rules to be followed for regions resulting from the image segmentation can be stated as: • They should be uniform and homogeneous with respect to some characteristics.

• Their interiors should be simple and without many small holes.

• Adjacent regions should have significantly different values with respect to the characteristic on which they are uniform.

• Boundaries of each segment should be simple, not ragged, and must be spatially accurate.

Various different colour spaces have been defined which simply described the colours, or gamut that particular electronic equipment can interpret, analyze or display. The choice of colour space representation could be taken to enhance the performance of processes such as segmentation because of the increase in demand of the colour driven images as compared to gray scale images [6]-[7].

II. IMAGE SEGMENTATION

In this paper, the image segmentation is defined as an optimal segmentation obtained in a pure bottom-up fashion that provides the information necessary to initialize and constrain high-level segmentation methods. Although the details of primary segmentation methods will depend on the application domain, we require that they do not depend on a priori knowledge about the objects present in a particular scene or image specific parameter adjustments. These claims become realistic because we do not seek for a perfect segmentation result but rather for the best possible support for more intelligent methods to be applied afterwards. Unfortunately up to now there is no theory which defines the quality of a segmentation. Therefore we have to rely on some heuristic constraints which the primary segmentation should meet:

 The segmentation should provide regions that are homogenous with respect to one or more properties, i.e. the variation of measurements within the regions should be considerably less than the variation at borders.

 The position of the borders should coincide with local maxima, ridges and saddle points of the local gradient of the measurements.

International Journal of Emerging Technology and Advanced Engineering

157

shape or contrast, should not be merged with their neighbouring regions.[10]

III. IMAGE PRE-PROCESSING

Image pre-processing is form of signal processing for which the input is an image, such as a picture; the output of image pre-processing may be either an image or, a set of characteristics or parameters related to the image. Most image pre-processing techniques involve treating the image as a two-dimensional signal and applying standard signal-processing techniques to it. segmentation refers to the process of partitioning a digital image into multiple segments (sets of pixels, also known as super pixels). The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze. Image segmentation is typically used to locate objects and boundaries in images.[9]

The two types of methods used for Image Processing are

 Analog Image Processing

 Digital Image Processing.

Analog or visual techniques of image processing can be used for the hard copies like printouts and photographs. Association is another important tool in image processing through visual techniques. So analysts apply a combination of personal knowledge and collateral data to image processing.

Digital Processing techniques help in manipulation of the digital images by using computers. As raw data from imaging sensors from satellite platform .The three general phases that all types of data have to undergo while using digital technique are Pre-processing, enhancement and display, information extraction.

A.Purpose Of Image Processing

The purpose of image processing is divided into 5 groups. They are:

 Visualization - Observe the objects that are not visible.

 Image sharpening and restoration - To create a better image.

 Image retrieval - Seek for the image of interest.

 Measurement of pattern – Measures various objects in a image.

 Image Recognition – Distinguish the objects in an image.

The pre-processing of the images. Pre- processing consists of those operations that prepare data for subsequent analysis that attempts to correct for systematic errors. The digital images are subjected to several corrections. After the pre-processing is complete, the original images are pre-processed to make the dimensionality more adaptable to processing which also helps to make the processing faster.

IV. LAB COLOUR SPACE

A Lab colour space is a colour opponent space with dimension L for lightness and a and b for the colour-opponent dimensions, based on nonlinearly compressed CIE XYZ colour space coordinates. "Lab" colour spaces is to create a space which can be computed via simple formulas from the XYZ space, but is more perceptually uniform than XYZ. Perceptually uniform means that a change of the same amount in a colour value should produce a change of about the same visual importance. When storing colours in limited precision values, this can improve the reproduction of tones. Both Lab spaces are relative to the white point of the XYZ data they were converted from. Lab values do not define absolute colours unless the white point is also specified.[11].Your goal is to identify different colours in image by analyzing the L*a*b* colour space. The image was acquired using the Image Acquisition Toolbox.

Step 1: Acquire Image

Read the image, which is an colourful image instead of using gray image.

Step 2: Calculate Sample Colours in L*a*b* Colour Space for each region.

The L*a*b* colour space is derived from the CIE XY tristimulus values. The L*a*b* space consists of a luminosity 'L*' layer, chromaticity layer 'a*' indicating where colour falls along the red-green axis, and chromaticity layer 'b*' indicating where

the colour falls along the blue-yellow axis. Your approach is to choose a small sample region for

each colour and to calculate each sample region's average colour in 'a*b*' space.

Step 3: Classify Each Pixel Using the Nearest Neighbour rule each colour marker now has an 'a*' and a 'b*' value.

International Journal of Emerging Technology and Advanced Engineering

158

Step 4: Display Results of Nearest Neighbour Classification The label matrix contains a colour label for each pixel in the fabric image. Use the label matrix to separate objects in the original fabric image by colour.

Step 5: Display 'a*' and 'b*' Values of the Labelled Colours. The nearest neighbour classification separated the different colour populations by plotting the 'a*' and 'b*' values of pixels that were classified into separate colours. For display purposes, label each point with its colour label.

Segmented Image

Figure1. Scheme of the segmentation method

V. DIFFERENT CHANNEL IN LAB COLOUR SPACE

The three coordinates of LAB represent the lightness of the color (L* = 0 yields black and L* = 100 indicates diffuse white; specular white may be higher), its position between red/magenta and green (a*, negative values indicate green while positive values indicate magenta) and its position between yellow and blue (b*, negative values indicate blue and positive values indicate yellow) coordinate ranges from 0 to 100.

The possible range of a* and b* coordinates is independent of the colour space that one is converting from, since the conversion uses X and Y which come from RGB the red/green and yellow/blue opponent channels are computed as differences of lightness transformations of cone responses, CIELAB is a chromatic value colour space

The nonlinear relations for L*, a*, and b* are intended to mimic the nonlinear response of the eye. Furthermore, uniform changes of components in the L*a*b* colour space aim to correspond to uniform changes in perceived colour, so the relative perceptual differences between any two colours in L*a*b* can be approximated by treating each colour as a point in a three dimensional space.

The L*a*b* colour space includes all perceivable colours which means that its gamut exceeds those of the RGB and CMYK colour models. One of the most important attributes

Figure2. The L*a*b* model

of the L*a*b*-model is the device independency. This

means that the colours are defined independent of their nature of creation or the device they are displayed on. The L*a*b* color space is used e.g. in Adobe Photoshop when graphics for print have to be converted from RGB to CMYK, Your goal is to identify different colours in image by analyzing the L*a*b* colour space

Image pre-processing

RGB Colour Space to L*a*b* colour

space ange

Channel Separation representing the different

Colours

International Journal of Emerging Technology and Advanced Engineering

159

Figure3. The CIE 1976 (L*, a*, b*) colour space

The next implementation in the proposed method is to convert the pre-processed images which are in RGB colour space to L*a*b* colour space. For this proposed work L*a*b* colour space is selected which is a homogeneous space for visual perception.

Figure4. Pre- Processed Image

The difference between the two points in the L*a*b* colour space is same with the human visual system. Since the L*a*b* model is a three-dimensional model, it can only be represented properly in a three-dimensional space [8]-[9]. The solution to convert digital images from the RGB space to the L*a*b* colour space is given by the following formula [8].

L* = 116 f(Y/Yn) – 16

a* = 500[f(X/Xn)-f(Y/Yn)]

b* = 200[f(Y/Yn)-f(Z/Zn)]

X, Y, Z, Xn, Yn, and Zn are the coordinates of CIEXYZ colour space. The solution to convert digital images from the RGB space to the CIEXYZ colour space is as the following formula.

X 0.608 0.174 0.201 R

Y = 0.299 0.587 0.114 G

International Journal of Emerging Technology and Advanced Engineering

160

f(x) = X1/3 x>0.008856

7.787x+16/116 x<= 0.008856

Colour space conversion is the translation of the representation of a colour from one basis to another. This typically occurs in the context of converting an image that is represented in one colour space to another colour space. The results after colour space transformation are shown below in fig3.

Figure5. L*a*b* Colour Space converted image

VI. PROPOSED ALGORITHM/METHOD

Genetic Algorithm (GA) is a population-based stochastic search procedure to find exact or approximate solutions to optimization and search problems. Modelled on the mechanisms of evolution and natural genetics, genetic algorithms provide an alternative to traditional optimization techniques by using directed random searches to locate optimal solutions in multi model landscapes [13,14]. Each chromosome in the population is a potential solution to the problem. Genetic Algorithm creates a sequence of populations for each successive generation by using a selection mechanism and uses operators such as crossover and mutation as principal search mechanisms - the aim of the algorithm being to optimize a given objective or fitness function. An encoding mechanism maps each potential solution to a chromosome. An objective function or fitness function is used to evaluate the ability of each chromosome to provide a satisfactory solution to the problem.

The selection procedure, modeled on nature’s survival-of-the-fittest mechanism, ensure that the fitter chromosomes have a greater number of offspring in the subsequent generations.

For crossover, two chromosomes are randomly chosen from the population. Assuming the length of the chromosome to be l, this process randomly chooses a point between 1 and l-1 and swaps the content of the two chromosomes beyond the crossover point to obtain the offspring. A crossover between a pair of chromosomes is affected only if they satisfy the crossover probability.

Mutation is the second operator, after crossover, which is used for randomizing the search. Mutation involves altering the content of the chromosomes at a randomly selected position in the chromosome, after determining if the chromosome satisfies the mutation probability. In order to terminate the execution of GA we specify a stopping criterion. Specifying the number of iterations of the generational cycle is one common technique of achieving this end.

A.Genetic Algorithm Based Clustering’s

The searching capability of GAs can be used for the purpose of appropriately clustering a set of n unlabeled points in N-dimension into K clusters [1]. In our proposed scheme, the same idea can be applied on image data. We consider a colour image of size mxn and every pixel has Red, Green and Blue components. The basic steps of the GA-clustering algorithm for clustering image data are as follows:

1. Encoding

Each chromosome represents a solution which is a sequence of K cluster centers. For an N dimensional space, each cluster center is mapped to N consecutive genes in the chromosome. For image datasets each gene is an integer representing an intensity value of the three components Red, Green and Blue.

2. Population initialization

Each of the P chromosomes is initialized to K randomly chosen points from the dataset. Here P is the population size.

3. Fitness computation

International Journal of Emerging Technology and Advanced Engineering

161

if || Xi(r,g,b)-Zj(r,g,b)|| < || Xi(r,g,b)-Zp(r,g,b)||, p = 1,2,...,K, and p ≠ j.

B.Proposed Method

We proposed a new segmentation algorithm that can produce a new result according to the values of the clustering. We consider a colour image f of size mxn.

The proposed algorithm is:

1. Repeat steps 2 to 8 for K=2 to K=Kmax. 2. Initialize the P chromosomes of the population. 3. Compute the fitness function fi for i=1,…,P, using

equation.

4. Preserve the best (fittest) chromosome seen till that generation.

5. Apply selection on the population.

6. Apply crossover and mutation on the selected population.

7. Repeat steps 3 to 6 till termination condition is reached.

8. Compute the clustering Validity Index for the fittest chromosome for the particular value of K.

9. Cluster the dataset using the most appropriate number of clusters determined by comparing the Validity Indices of the proposed clusters for K=2 to K=Kmax.

VII. EXPECTED RESULT

The result of image segmentation is a set of segments that collectively cover the entire image, or a set of contours extracted from the image. Each of the pixels in a region are similar with respect to some characteristic or computed property, such as colour, intensity ,or texture. Adjacent regions are significantly different with respect to the same characteristic. When applied to a stack of images, typical in Medical imaging, the expected result contours after image segmentation can be used to find how many objects in cluster as well as it is used to count a regions in image.

VIII.CONCLUSION

A new colour image segmentation method is proposed, which utilizes the general method. The mathematics of the proposed method is discussed comprehensively and expected results are presented. Comparison of the performance of the proposed method with an available clustering method ,I expect that the proposed method is more stable and faster.

It is also observed that the proposed method decreases the probability of local minimum entrapment. The usability of the proposed segmentation method is also more than the available methods. Furthermore, while the proposed method gives more perceptually satisfactory segmentation results, it demands less processing resources. the concept of segmentation based on the colour features of an image.

IX. FUTURE ENCHANCEMENT

A new feature selection technique for face recognition we can proposed. the most proper ones should be selected to enhance the performance of classification. Although GA considered as one the best optimization methods, defining an appropriate and global fitness function for the feature selection has a high impact on its performance. The associated problem of simple GA fitness function was the quick convergence into uninformative feature sets. In the proposed techniques named Swap Training a new fitness function .

REFERENCES

[1 ] Kearney,Colm and Patton, J. Andrew, “Survey on the image segmentation”, Financial Review, 41: 29-48 (2000).

[2 ] H. D. Cheng, X. H. Jiang, Y. Sun, and J. Wang, “Color image segmentation: advances and prospects,” Pattern Recognition, 34: 2259–2281, (2001).

[3 ] He, Xiaoling; Hodgson, W. Jeffrey ,”Research Image Processing Technology hot issue”, IEEE Transactions on Intelligent Transportation Systems, 3(4): Dec., 244-251 (2002).

[4 ] Xia Yong and Feng Dagan; "A General Image Segmentation Model and its Application," icig , pp.227-231, 2009 Fifth International Conference on Image and Graphics, 2009.

[5 ] Y. Boykov and V. Kolmogorov; “An experimental comparison of min-cut/max-flow algorithms for energy minimization in vision”,IEEE Trans. On Pattern Analysis and Machine Intelligence, vol. 26, no. 9, pp. 1124-1137, 2004.

[6 ] Kwok, M.N. ; Ha, P.Q. and Fang, G. - Image and Signal Processing “Effect of color space on color image segmentation”.In: Image and Signal Processing. CISP ’09. 2nd International Congress on ,17-19 Oct., Tianjin (2009).

[7 ] Erik Reinhard, Michael Adhikhmin, Bruce Cooch, et al. ;“Color Transfer between Image”, IEEE transactions on Computer Graphics and Applications, USA, 2001.

[8 ] Chun Chen;“Computer image processing technology and algorithms”., Beijing: Tsinghua University Press, 2003.

[9 ] R. Fisher, K Dawson-Howe, A. Fitzgibbon, C. Robertson, E. Trucco (2005). Dictionary of Computer Vision and Image Processing. John Wiley. ISBN 0-470-01526-8

International Journal of Emerging Technology and Advanced Engineering

162

Conundrum and Other Adventures in the Most Powerful Colorspace. Berkeley, Calif. : London: Peachpit ; Pearson Education. ISBN 0321356780.

[12 ]COLORLAB MATLAB toolbox for color science computation and accurate color reproduction. It includes CIE standard tristimulus colorimetry and transformations to a number of non-linear color appearance models (CIE Lab, CIE CAM, etc.).

[13 ]M. Srinivas, Lalit M. Patnaik, “Genetic Algorithms: A urvey”. [14 ]D. E. Goldberg, Genetic Algorithms in Search, Optimization and

Machine Learning, Addison-Wesley, 1989.