HYBRID K-MEANS CLUSTERING FOR COLOR IMAGE SEGMENTATION

2017

Segmentation implies the division of an image into different objects or connected regions that do not overlap. Though, extensive research has been done in creating many different approaches and algorithms for image segmentation, however, it is still not very clear to assess whether one algorithm produces more accurate segmentations than another, whether it be for a particular image or set of images, or more generally, for a whole class of images [7]. A reliable and accurate segmentation of an image is, in general, very difficult to achieve by purely automatic means. Present researches on image segmentation using clustering algorithms reveals that K-means clustering algorithm so far produces best results but some improvements can be made to improve the results. The biggest disadvantage of our heavy usage of k-means clustering, is that it means we would have to think of a k each time, which really doesn’t make too much sense because we would like to algorithm to solve this on his own....

Image segmentation divides an image into several constituent components such as color, structure, shape, and texture. It forms a major research topic for many image processing researchers as the applications are endless. Its applications include image enhancement, object detection, image retrieval, image compression, and medical image processing to name a few. The segmentation of color images is necessary for efficient pattern recognition and feature extraction involving various color spaces such as RGB, HSV and CIE L*A*B*, etc. This paper describes the different cluster based segmentation techniques used for segmenting the different color images and the resultant is analyzed with subjective and objective measures. Initially, registered color images are considered as input. Then the cluster based segmentation techniques namely K-Means clustering, Pillar-Kmeans clustering and Fuzzy C-means (FCM) clustering techniques are applied. Further, the segmented image is analyzed with measures such as compactness and execution time. From the experimental results, it has been observed that K-means and Pillar-Kmeans are the most suitable techniques for RGB, HSV and LAB color spaces than the FCM technique.

Color image segmentation is currently a very emerging topic for researchers in Image processing. Clustering is a frequently chosen methodology for this image segmentation task. But for a better segmentation, there arises the need of an optimal technique. In this paper, we propose an integrated approach for color image segmentation which is a new of its kind. Here, we integrate the famous k-means algorithm with watershed algorithm. But, here we chose ‘cosine’ distance measure for k-means algorithm to optimize the segmented result of the later one. Also, as color space has a leading impact on color image segmentation task, so, we chose HSV color space for our proposed approach. Since usually the noise arises during the segmentation process, so here the final segmented image is filtered by median filter to make the output image clearer and noise free. The result of the proposed approach is found to be quite satisfactory.

2016

A comparative analysis of color image segmentation of three different techniques is presented in this paper with different noise levels. The robustness of each algorithm is checked on the basis of its accuracy of the clusters mapped. We have taken K-Means, FCM and Density based image segmentation approach for the analysis. In these methods of segmentation, the objects are distinguished clearly from the background. Basically Image is that type of information which has to be processed effectively and correctly. Segmentation of an image requires the separation or division of the image into regions of similar and multiple attribute. Image segmentation assigns a tag to each and every pixel in an image such that each pixel with the same tag share certain visual characteristic. Basic attribute for segmentation of an image is its luminance amplitude for an image and color components for a color image and its intensity level. Clustering is one of the methods which are used for segmentation. ...

INTERNATIONAL JOURNAL OF COMPUTERS & TECHNOLOGY

Very large collections of images are growing ever more common. From stock photo collections to proprietary databases to the Web, these collections are diverse and often poorly indexed; unfortunately, image retrieval systems have not kept pace with the collections they are searching. The shortcomings of these systems are due both to the image representations they use and to their methods of accessing those representations to find images: (i) When users generally want to find images containing particular objects, most existing image retrieval systems represent images based only on their low-level features (“stuff”) with little regard for the spatial organization of those features. (ii) Systems based on user querying are often unintuitive and offer little help in understanding why certain images were returned and how to refine the query. Often the user knows only that he has submitted a query for, say, a bear and retrieved very few pictures of bears in return. (iii) For general ima...

Color is one of the properties which add information to the images. Classes of pixels are difficult to be identified when the color distributions of the different objects highly overlap in the color space and when the color points give rise to non-convex clusters. Color based image segmentation using fuzzy c means and k means algorithms can be used for the clustering of color image. This method is used to cluster and measure accuracy of the color images by segmenting each color pixels in the color images. Once segmentation is done, the fuzzy c means method is used for creating membership operation functions to define the degree to which a pixel belongs to an edge or a uniform region. The k-means clustering is used to partition n data points into k clusters. This unsupervised clustering approaches has a strong affinity to get trapped into local minima while generating an optimal solution. Hence, it makes clustering wholly dependent on the distribution of primary cluster centre. This research work is employed to find the distance between color pixels of the RGB color spaces. Implementation has been done using MATLAB Simulation tool which generates the better result of this clustering algorithms.

International Journal of Applied Evolutionary Computation, 2015

The findings of image segmentation reflects its expansive applications and existence in the field of digital image processing, so it has been addressed by many researchers in numerous disciplines. It has a crucial impact on the overall performance of the intended scheme. The goal of image segmentation is to assign every image pixels into their respective sections that share a common visual characteristic. In this paper, the authors have evaluated the performances of three different clustering algorithms normally used in image segmentation – the typical K-Means, its modified K-Means++ and their proposed Enhanced Clustering method. The idea is to present a brief explanation of the fundamental working principles implicated in these methods. They have analyzed the performance criterion which affects the outcome of segmentation by considering two vital quality measures namely – Structural Content (SC) and Root Mean Square Error (RMSE) as suggested by Jaskirat et al., (2012). Experimental...

In this paper color based image segmentation is done in two spaces. First in LAB color space and second in RGB space all that done using three versions of K-Means: K-Means, Weighted K-Means and Inverse Weighted K-Means clustering algorithms for different types of images: biological images (tissues and blood cells) and ordinary full colored images. Comparison and analysis are done between these three algorithms in order to differentiate between them.

INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY, 2014

This paper present image segmentation using Enhanced k-means clustering with divide and conquer approach. First we enhanced the k-means clustering and then segment the image using enhanced approach. K-means is one of the most popular clustering algorithms. The final clustering result of the k-means clustering algorithm greatly depends upon the correctness of the initial centroids, which are selected randomly. Many improvements were already proposed to improve the performance of k-means, but most of these require initial k centroids .In this paper we have proposed a new method enhanced k-means clustering algorithm with divided and conquer algorithm to find the initial centroids with reduced time complexity. In this paper we fixed the initial centroids which gives best result and in the final step we segment the image using enhanced clustering algorithm.

IJRCAR, 2014

Image processing is an important research area in computer vision. Image segmentation plays the vital rule in image processing research. There exist so many methods for image segmentation. Clustering is an unsupervised study. Clustering can also be used for image segmentation. In this paper, an in-depth study is done on different clustering techniques that can be used for image segmentation with their pros and cons. An experiment for color image segmentation based on clustering with K-Means algorithm is performed to observe the accuracy of clustering technique for the segmentation purpose.