Loading...
Please wait, while we are loading the content...
Similar Documents
Analysis of Color Image Segmentation Using Cluster Based Self-Organizing Map (SOM) Algorithm
| Content Provider | Semantic Scholar |
|---|---|
| Author | Soe, Thida |
| Copyright Year | 2019 |
| Abstract | Image segmentation is an important role in digital image processing and it could be solved by many clustering method. Segmentation of an image entails the division or separation of an image into regions of similar attribute. In this proposed system, initially natural images are taken from the Berkley Image Segmentation Database (BSD). Various color space of images such as RGB, HSV and L* A* B* are used as input images for the segmentation process. In order to get the same size of image images with different color space, Image J software is used to in this system. Because color conversion function may reduce the input images size is not flexible for this system. This system uses the different color images and the resultant is analyzed with subjective and objective measures. Then the cluster based segmentation techniques SOM unsupervised clustering techniques is applied. A self-organizing map (SOM) is a type of artificial neural network (ANN) that is trained using unsupervised learning to produce a low-dimensional input space of the training samples. This developed method takes into account the color similarity and spatial relationship of objects within an image. According to the features of color similarity, an image is first segmented into cluster regions. The resulting regions are further treated by computing the spatial distance between any two cluster regions, and SOM with a labeling process is applied. The experimental results show that the proposed system is feasible and that the segmented object regions are similar to those perceived by human vision. KeywordsColor Image Segmentation, SOM Algorithm, Image Processing, Color Space Model, ANN 1. INTRDUCTION Color image segmentation is the challenging task in image processing and contains two critical issues, firstly which color model to be used and secondly, which segmentation technique should be applied. A color space is a method by which we can specify, create and visualize colors. Several color representations, such as RGB, HSI, CMY, CMYK, YIQ, CIE L∗a∗b∗,etc., are employed for color segmentation, but none of them can dominate the others for all kinds of colors images. Image segmentation is the process of partitioning a digital image into multiple segments. Each segment will represent some kind of information to user in the form of color, intensity or texture. The goal of segmentation is to simplify or change the representation of an image into that is more meaningful and easier to analyze. Image segmentation is a useful tool in many real system including industry, health care, astronomy and remote sensing, biomedical imaging, change detection, object detection and recognition. Clustering methods provide with a different view of the image segmentation by using different color spaces with same size of input images. SOM studies each inputs component and then classifies the input into the corresponding class. The K-Means clustering technique is a well-known approach that has been applied to solve low-level image segmentation tasks. A successful segmentation depends on the good selections of similarity measure, feature description of an image, evaluation of the segmentation and prior knowledge available. This clustering algorithm is convergent and its aim is to optimize the partitioning decisions based on a user-defined initial set of clusters that is updated after each iteration. This work was motivated from the fact that the accuracy in segmentation of a color image depends not only on the algorithm but also on the color space selected. K means clustering algorithm which involves mapping the image pixel to the RGB color space and HSV color space [1]. HSV color space will be more compatible for conduct with segmentation of rough color images. RGB color space is also called additive color space, which can be described well based on the RGB color model. In [2], three different cluster based segmentation techniques performed on 3 different color images .But in K-means technique we can get various segments according to cluster size. K-Means provided better results than the other two techniques using subjective (visualization and execution time) measure for RGB, HSV and LAB color spaces. In [3] , K-Means, Fuzzy C-Means and Density Based clustering techniques are compared for their performance in segmentation of color images. K-Means algorithm is convergent and its aim is to optimize the partitioning decisions based on a user-defined initial set of clusters. Fuzzy C-means (FCM) is a method of clustering which allows one pixel to belong to two or more clusters. Image segmentation based on density-based clustering; will integrate the spatial connectivity and the color similarity simultaneously in the segmentation process. Using these three techniques, the performance for different images were evaluated by calculating their accuracy. Self-Organizing Map (SOM) is an unsupervised artificial neural network technique that is used to produce a low dimensional representation of the input space [4]. SOM operates in two modes, Training and Mapping. Training builds the map from the input and Mapping classifies a new input vector. Selforganizing maps are capable of maintaining spatial information of the image and are therefore preferred whenever preserving the topological features is a priority in this paper. It is show that the proposed K-Means algorithm, gets better segmentation results with less time needed and no need to set any parameters in advance. © 2019 JETIR January 2019, Volume 6, Issue 1 www.jetir.org (ISSN-2349-5162) JETIR1901A82 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 676 2. COLOR IMAGE SEGMENTATION Clustering is the task of partitioning the data points into homogeneous classes or clusters so that items in the same class are as similar as possible and items in different classes are as dissimilar as possible. Clustering can also be thought of as a form of data compression, where a large number of samples are converted into a small number of representative prototypes or clusters. Depending on the data and the application, different types of similarity measures may be used to identify classes, where the similarity measures controls how the clusters are formed. As a kind of unsupervised learning method, clustering is divided to be a hierarchical and partition. 2.1 Clustering Based Segmentation Clustering is the task of partitioning the data points into homogeneous classes or clusters so that items in the same class are as similar as possible and items in different classes are as dissimilar as possible. Clustering can also be thought of as a form of data compression, where a large number of samples are converted into a small number of representative prototypes or clusters. Depending on the data and the application, different types of similarity measures may be used to identify classes, where the similarity measures controls how the clusters are formed. As a kind of unsupervised learning method, clustering is divided to be a hierarchical and partition. SOM differ from other artificial neural networks as they apply competitive learning as opposed to error-correction learning and they use as a neighborhood function to preserve the topological properties of the input space. Being a typical partitioned clustering method, K-Means method assign each points to the cluster with the nearest center. 2.2 Color Space Model A color space is a useful method for user to understand the color capabilities of a particular digital device or file. An RGB color space can be simply interpreted as “all possible colors” which can be made from three colors for red, green and blue. The HSV color space is used when selecting colors for paint or ink because HSV better represents how people relate to colors than does the RGB color space. Selecting an HSV color begins with picking one of the available hues, which adjusting the shade and brightness value. LAB color is designed to approximate human vision. L component closely matches human perception of lightness. It can be used to make accurate color balance corrections by modifying output in a and b components, or to adjust the lightness contrast using the L component. The system use Image J tools combine with color conversion plugin for color space conversion like that RGB to HSV and RGB to LAB. This tool designed with an open architecture that provides extensibility via Java plugins. User-written plugins make it possible to solve almost any image processing or analysis problem. It supports standard image processing functions such as contrast manipulation, sharpening, smoothing, edge detection and median filtering. It can display, edit, analyze, process, save and print 8-bit, 16-bit and 32-bit images. It can read many image formats including TIFF, GIF, JPEG, PNG, DICOM, BMP, PGM, FITS and so on. 3. Proposed System Architecture This proposed system is to simplify and change the representation of an image into something that is more meaningful and easier to analyze. Self-Organizing Map (SOM) for color clustering which is applies in color image segmentation process. Learning SOM algorithm for training with neural network structure and feature of prototype vector for color image segmentation. In this experiment, the effectiveness of low clustering methods involving RGB, HSV, L*a*b color spaces for a variety of real color image is obtained. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.jetir.org/papers/JETIR1901A82.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
