Int. J. Advanced Networking and Applications Volume: 6 Issue: 2 Pages: 2222-2232 (2014) ISSN : 0975-0290 2222 Applying Content-Based Image Retrieval Techniques to Provide New Services for Tourism Industry Zobeir Raisi Department of Electrical Engineering, Chabahar Maritime University, Chabahar, IRAN Email: zobeir.raisi@cmu.ac.ir Farahnaz Mohanna Assistant professor, Faculty of Electrical and Computer Engineering, University of Sistan and Baluchestan, Zahedan, IRAN Email: f_mohanna@ece.usb.ac.ir Mehdi Rezaei Assistant professor, Faculty of Electrical and Computer Engineering, University of Sistan and Baluchestan, Zahedan, IRAN Email: mehdi.rezaei@ece.usb.ac.ir ----------------------------------------------------------------------ABSTRACT-------------------------------------------------------------- The aim of this paper is to use the network and internet and also to apply the content based image retrieval techniques to provide new services for tourism industry. The assumption is a tourist faces an interesting subject; he or she can take an image of subject by a handheld device and send it to the server as query image of CBIR. In the server, images similar to the query are retrieved and results are returned to the handheld device to be shown on a web browser. Then, the tourist can access the useful information about the subject by clicking on one of the retrieved images. For this purpose, a tourism database is created. Then several particular content-based image retrieval techniques are selected and applied to the database. Among these techniques, ‘Edge Histogram Descriptor (EHD)’ and ‘Color layout descriptor (CLD)’ algorithms have better retrieval performances than the others. By combining and modification of these two methods, a new CBIR algorithm is proposed for this application. Simulation results show a high retrieval performance for the proposed algorithm. Keywords – CBIR, Tourism Industry, MPEG-7, Mobile, Handheld Device ----------------------------------------------------------------------------------------------------------------------------------------------------- Date of Submission : August 06, 2014 Date of Acceptance : September 29,2014 ----------------------------------------------------------------------------------------------------------------------------------------------------- 1. INTRODUCTION Nowadays, with the explosive growth of the number of digital images available on internet and the availability of image capturing devices such as digital cameras and image scanners, the size of digital products is increasing rapidly [1, 2], therefore efficient indexing and searching becomes essential for large image archives. For this purpose, many general-purpose image retrieval systems have been developed. There are three categories of image retrieval methods: text-based, content-based, and semantic based. In text-based systems, the images are manually annotated by text descriptors. Text annotation to all images manually is impractical because of large labeling cost and the subjective of human perception. To overcome the above disadvantages in text-based image retrieval system, ‘Content-Based Image Retrieval (CBIR)’ was introduced in the early 1980s, which is based on automatically indexing and retrieval [3, 4]. CBIR aims to search images that are perceptually similar to the query image based on visual content of the images without the help of annotations. Researches mainly focused on the effective low-level representation of images and CBIR usually indexes images by low-level visual features such as color [5], texture [6], and shape [7]. Color is the most dominant and distinguishing visual feature that is widely used in CBIR and is invariant to image size and orientations [8, 9]. As conventional color features used in CBIR, there are color histogram, color correlogram, color structure descriptor, and scale color descriptor [2]. Color histogram is the most commonly used color representation scheme to represent the global feature composition of an image but it does not have any spatial information. It is invariant to translation and rotation of an image and normalizing the histogram lead to scale invariance [10]. Texture is used to specify the roughness or coarseness of object surface and described as a pattern with some kind of regularity. Texture feature has been used in various applications ranging from industrial application to medical imaging. There are various algorithms for texture analysis used by researches, such as gray co-occurrences matrix [11], Markov random field [12], ‘simultaneous auto-regressive (SAR)’ model [13], wold decomposition model [14], Gabor filtering [15], wavelet decomposition [16] and so on[7, 8]. Shape features are important image features though they have not been widely used in CBIR as color and texture features. Accurate extraction and representation of shape information is one of the challenging tasks in shape image retrieval [17]. Shape features have shown to be useful in many domain specific images such as man-made objects. For color images used in most papers, however, it is difficult to apply shape features compared to color and texture due to the inaccuracy of segmentation [7]. The classic methods of