International Journal of Computer and Communication Engineering, Vol. 1, No. 4, November 2012 317 Abstract—This paper presents an enhancement of the performance of image clustering. K-Means has been chosen as our clustering technique. We applied Hue Saturation Value (HSV) color histogram as features to retrieve image information. Singular Value Decomposition (SVD) technique is employed to enhance the performance of image clustering by reducing features that not useful to be proceed in clustering process. We evaluated the image clustering using Recall, Precision, F-Measure and computational time. The obtained result indicates that SVD in HSV color histogram Content-Based Image Retrieval (CBIR) is promising. Index Terms—Content based image retrieval, HSV color histogram, k-means algorithm, singular value decomposition I. INTRODUCTION Currently the increasing of image data on internet is giving opportunity for researcher to working in searching and classifying the content of image. The example of image management tool that capable to search and classifying image is Google image as image searching application usually used for image mining and image search in web context. The importance of searching image information in web context given many research communities has produced many method algorithms with tools for image retrieval and clustering, where this technique also include in technique Content-Based Image Retrieval (CBIR). There are many applications of CBIR such as biomedicine, military, commerce, education, web image classification and searching [1]. CBIR is a technique to searching by analyzes the actual content (feature) of image. In CBIR there are two techniques that should strong to achieve accurate results; there are a technique to retrieve information and a technique to cluster for classifying image. Image retrieval is a process to retrieve image features that include in each image. In [2] has stated two types of images features there are low level feature and high level feature, where high level feature is difficult to extract like emotion or other human behavior activities. There are several image features at low level feature usually used to retrieve image information such as feature color, shape and texture. In this study we only used color feature to retrieve the information of image based on Hue Saturation Value (HSV) space format. Manuscript received July 24, 2012; revised September 2, 2012. The authors are with the Dept. of Computer Science Dian Nuswantoro University, Semarang, Indonesia (e-mail: caturs@research.dinus.ac.id, guruh.fajar@research.dinus.ac.id, ricardus.anggi@research.dinus.ac.id, pulung@research.dinus.ac.id). This research applied K-Means as clustering technique, where this method commonly used for partitioning [3], [4]. Each cluster in K-Means method is represented by its centroids or the mean value of all data in the cluster. We applied color histogram technique that usually used to retrieve information image. The weakness applicable of this features extraction for information retrieval has been state by [5] where the implementation of color histogram does not give relevant image as seen by an algorithm with human visual. The aim of this paper is to adapt Singular Value Decomposition (SVD) as feature transformation in CBIR. SVD is a Latent Semantic Analysis (LSA) approach. The reason of using SVD is provide useful information of the color. CBIR system can be improved by using image clustering [6]. The problem of image clustering is high number of image in clustering leads to more computational time [7]. To overcome this problem, SVD also can be proposed to reduce the computational time of image clustering. The outline of this paper is as follows: section 2 describes the related work. Section 3 describes the methodology of research. Section 4 describes the dataset and shows the performance analysis of proposed approach. Section 5 presents the conclusion and future work. II. RELATED WORK There are several research has been conduct to improve the performance of CBIR in image retrieval clustering area. The previous work in [8]-[10] have been analyzed the performance of clustering algorithm for image retrieval. Kucuktunc and Zamalieva [11] proposed fuzzy color histogram for CBIR. Mamdani fuzzy inference system was used as fuzzy technique to link L*a*b* to fuzzy color space. Their work show that fuzzy color histogram performed better than conventional methods. Other study in [12] was compare the using of Conventional Color Histogram (CCH), Invariant Color Histogram (ICH) and Fuzzy Color Histogram (FCH) of an images in CBIR system. ICH and FCH has been used to address the problem of rotation, translation and spatial relationship of ICH. Tonge [13] proposed a K-means clustering algorithm to grouping the collection of images. Image is clustered based on the query image. Color was used to be features in their CBIR system. Sakthivel et al. [14] proposed modified K-Means clustering to group similar pixel in CBIR. Their purpose is to improve retrieval performance by capturing the regions and Performance Enhancement of Image Clustering Using Singular Value Decomposition in Color Histogram Content-Based Image Retrieval Catur Supriyanto, Guruh Fajar Shidik, Ricardus Anggi Pramunendar, and Pulung Nurtantio Andono