In our system we used Wikipedia and we created a resource basedon categories that we found here. To accomplish this, we started with Ro- manian Wikipedia which has 8 groups of categories: culture, geography, history, mathematics, society, science, technology, privacy. In turn, these categories have subcategories or links to pages directly, as follows: Culture (30) (among which we mention photo, architecture, art, sports, tourism, etc.) Geography (15) (among which mention Romania, Africa, Europe Countries, maps, etc.), History (6) (among which mention After the recall, By region, etc.), Mathematics (11) (among which mention Algebra, Arithmetic, Economics, Geometry, Logic, etc.), Society (22) (among which mention Anthropology, Archaeology, Business, Commu- nications, Philosophy, Politics, etc.), Science (23) (among which men- tion Anthropology, Archaeology, Astronomy, Biology, etc.), Technology (19) (among which mention Agriculture, Architecture, Biotechnology, Computer, etc.), Private life (8) (among which mention the Fireplace, Fun, People, Health, etc.). In the end, we obtained 8 big groups with 134 categories, which are subdivided into several subcategories and pages (hierarchical depth depends on each category and subcategory). In general, this hierarchy covers most of the concepts available for Ro- manian. For example, for Sport, we obtained 70 subcategories contain- ing other subcategories and 9 pages. Going through these categories and subcategories, we built specific resources with words that signal concepts of type person, location and organization. Some examples of signal words from these categories are:
Several problems remain including retrieval of features basedon location within animage, the extension of 2-dimensional features to 3-dimension, and appropriate segmentation of video images. Although research in higher order CBIR is un- derway, current systems are not capable of retrieving all in- stances of horses basedon the shape, color, or texture of a single instance of a horse in a query. For example, a shape- based query depicting a side view of a horse does not retrieve images of horses from behind or above. Research in object recognition conducted by Forsythe et al (1997) has sought to develop techniques for modeling a class of objects and identi- fying defining attributes and features for that class. Rorvig has examined the use of human judgments to train the system to recognize patterns of user-defined similarity for automatic identification of image classes. Chang et al, (1998) also util- ized users' relevance judgments to refine searches and to as- sign semantic keywords to images that can be used by subse- quent users to query the system.
Abstract—We present a scalable imageretrievalsystembased jointly ontext annotations and visual content. Previous ap- proaches in content basedimageretrieval often suffer from the semantic gap problem and long retrieving time. The solution that we propose aims at resolving these two issues by indexing and retrieving images using both their text descriptions and visual content, such as features in colour, texture and shape. A query in this system consists of keywords, a sample imageand relevant parameters. The retrieving algorithm first selects a subset of images from the whole collection according to a comparison between the keywords and the text descriptions. Visual features extracted from the sample image are then compared with the extracted features of the images in the subset to select the closest. Because the features are represented by high-dimensional vectors, locality sensitive hashing is applied to the visual comparison to speedup the process. Experiments were performed on a collection of 1514 images. The timing results showed the potential of this solution to be scaled up to handle large image collections.
Imageretrieval is an active research area inimageprocessing, pattern recognition, and computer vision. For the purpose of effectively retrieving more similar images from the digital image databases, this paper uses the local HSV color and Gray level co-occurrence matrix (GLCM) texture features. The image is divided into sub blocks of equal size. Then the color and texture features of each sub-block are computed. Color of each sub-block is extracted by quantifying the HSV color space into non-equal intervals and the color feature is represented by cumulative color histogram. Texture of each sub-block is obtained by using gray level co-occurrence matrix. An integrated matching scheme basedon Most Similar Highest Priority (MSHP) principle is used to compare the query and target image. The adjacency matrix of a bipartite graph is formed using the sub-blocks of query and target image. This matrix is used for matching the images. Euclidean distance measure is used in retrieving the similar images. As the experimental results indicated, the proposed technique indeed outperforms other retrieval schemes in terms of average precision.
Content BasedImageRetrieval (CBIR) has become an important area of research with the ever increasing demand and use of digital images in various fields such as medicine, engineering, sciences, digital photography etc. Unlike the traditional method of text-basedimageretrievalin which the image search is basedon textual description associated with the images, CBIR systems retrieve images basedon the content of the image such as colour, texture, shape or any other information that can be automatically extracted from the image itself and using it as a criterion to retrieve content related images from the database. The retrieved images are then ranked according to the relevance between the query imageand images in the database in proportion to a similarity measure calculated from the features .
The second approach takes a different stand and treats images and texts as equivalent data. It attempts to discover the correlation between visual features and textual words onan unsupervised basis, by estimating the joint distribution of features and words and posing annotation as statistical inference in a graphical model. For example imageretrievalsystembasedon decision trees and rule induction was presented in  to annotate web image using combination of image feature and metadata, while in , a system that automatically integrate the keyword and visual features for web imageretrieval by using association rule mining technology. These approaches usually learn the keywords correlations according to the appearance of keywords in the web page, and the correlation may not reflect the real correlation for annotating Web images or semantic meaning of keywords such as synonym . Ontology-basedimageretrieval is an effective approach to bridge the semantic gap because it is more focused on capturing semantic content which has the potential to satisfy user requirements better [10,11]. While semantically rich ontology addresses the need for complete descriptions of imageretrievaland improves the precision of retrieval. However, the lack of text information which affects the performance of keyword approach is still a problem intext ontology approach. Ontology works better with the combination of image features .this paper presents a new framework for web imageretrieval search engine which relay not only on ontology to discover the semantic relationship between different keywords inside the web page but also propose a new voting annotation technique extract the shared semantically related keywords from different web pages to eliminate and solve the problem of subjectivity of image annotation of traditional approaches and enhance the performance of the retrieval results by taking the semantic of the correlated data into consideration.
Content BasedImageRetrieval (CBIR) includes a set of methods for processing visual features of a query image to find similar images at an images database. As extracting features and determining similarity measure, are two main stages inretrieval systems. In this paper we have tried to give weight to image pixels and suggest an effective feature vector by use of eye physiological structure and annotation issue. Then modify feature components weight and optimize similarity measure by use of information of relative and irrelative images in each feedback. So by means of annotation issue in physiological structure of eye, pixels in center of image are more important and have more effective role in extracted features and also system accuracy will be increased for determining similar images, by means of optimizing similarity measure in each stage.
Abstract —The construction of a high-resolution panoramic image from a sequence of input overlapping images of the same scene is called image stitching/mosaicing. It is considered as an important, challenging topic in computer vision, multimedia, and computer graphics. The quality of the mosaic imageand the time cost are the two primary parameters for measuring the stitching performance. Therefore, the main objective of this paper is to introduce a high-quality image stitching system with least computation time. First, we compare many different features detectors. We test Harris corner detector, SIFT, SURF, FAST, GoodFeaturesToTrack, MSER, and ORB techniques to measure the detection rate of the corrected keypoints andprocessing time. Second, we manipulate the implementation of different common categories of image blending methods to increase the quality of the stitching process. From experimental results, we conclude that ORB algorithm is the fastest, more accurate, and with higher performance. In addition, Exposure Compensation is the highest stitching quality blending method. Finally, we have generated animage stitching systembasedon ORB using Exposure Compensation blending method.
structure to describe its main components by means of the here named Newscast Discursive Metadata (NDM), as well as techniques from image analysis and data mining domains. The NDM describes aspects such as screen time and field size of newscasts’ participants and the theme addressed in each newscast component. The proposed approach was developed in partnership with the Brazilian TV channel Rede Minas inan attempt to provide media analysts with tools to assist their work, consisting in one of the components of an information systemand created to support the discourse analysis of television programs (Pereira et al., 2015). As far as we know, it is the first methodology and approach for dealing with this demand for TV newscasts. Another contribution of this work was to develop a new methodology for registration and indexing of multimedia content, implemented in a Web information system.
From The late 90 th , "Skin Detection" becomes one of the major problems inimageprocessing. If "Skin Detection" will be done in high accuracy, it can be used in many cases as face recognition, Human Tracking and etc. Until now so many methods were presented for solving this problem. In most of these methods, color space was used to extract feature vector for classifying pixels, but the most of them have not good accuracy in detecting types of skin. The proposed approach in this paper is basedon "Color basedimageretrieval" (CBIR) technique. In this method, first by means of CBIR method andimage tiling and considering the relation between pixel and its neighbors, a feature vector would be defined and then with using a training step, detecting the skin in the test stage. The result shows that the presenting approach, in addition to its high accuracy in detecting type of skin, has no sensitivity to illumination intensity and moving face orientation.
The Content-BasedImageRetrieval (CBIR) technique uses image content to search and retrieve digital images. Basically, CBIR systems try to retrieve images similar to a user-defined specification or pattern (e.g., shape sketch, image example). Their goal is to support imageretrievalbasedon content properties (e.g., shape, color, texture) , CBIR is also basedon the idea of extracting visual features from the imageand using them to index images in a database. Content- basedimageretrieval systems were introduced to address the problems associated with text-basedimageretrieval. CBIR is a set of techniques for retrieving semantically-relevant images from animage database basedon automatically-derived image features. Content-basedimageretrieval also known as query by image content is a technique which uses visual content that well known as features for extracting similar images from animagein a database. Image database every time become bigger and it makes a problem dealing with database organization so the necessity of efficient algorithm is obvious needed .
The first set of experiments is related to the problem of the analysis of NDT andtext images. NDT means to detect an object and quantify its possible defects without harmful effects on it by special equipments and methods . In this study, two real images are used in the experiments to assess the performance of the proposed method. The first image is NDT image employed in a recent work of Li et al. . It represents a light microscopy image of a material structure. Light microscopy is frequently used for inspecting the microstructures of materials to derive information about their properties such as the porosity, the particle sizes, the distribution uniformity, etc. The second image is a lenience plate image. Accurate segmentation of textin a plate lenience image plays a crucial role in a lenience plate recognition system. Obviously, the histogram of this gray level data is non-Gaussian in nature. Fig. 1 and Fig. 2 show that, in both ‘material image’ and ‘textimage’, the two classes are distinct in size, intensity range and skewness. The results of thresholds, indicate that the MET method has serious under-segmentation for textimageand conversely for material image. The performance of the FI-based extension of MET method is superior to that of the original MET method.
As discussed above, we can conclude that both TBIR and CBIR have their own characteristics, advantages and disadvantages. Low level visual features of animage represents the more detailed perceptual aspects while text addresses the high level semantics underlying the more general conceptual aspects of animage. Efforts have been made by the researchers to combine these two approaches to provide us with satisfactory results . In 1999 an important research work is done focusing on Content basedretrieval inspired from textretrieval . Abbas et.al.  suggests that combination of both textand content could better the performance of search systems benefitting both the approaches. It is based upon the idea if ability to examine the image content does not exist, then search depends upon metadata like captions/keywords.it says that TBIR is as fast as CBIR. An effort has been made to combine content and semantics in Medical domain also . The paper proposes a scheme to combine CBIR and semantics using grid computing The MedImGrid used in the proposed CSBIR extracts semantic context information and use it as clue to improve efficiency. The shape, color and texture features of every medical image for thorax CR based CSBIR is extracted and Euclidean distance metric is used for the similarity matching of visual content of the image. A recent study says content and context is required to bridge the gap rather than content vs. context . This paper gives four reasons to support this and present approaches that appropriately combine content with context to bridge the semantic gap. They suggested a new direction based upon philosophy, cognitive science and modern search engines that can be easier to bridge the semantic gap. C. Hartvedt  discusses how combining existing techniques may help improve the understanding of user intentions in IR. Underlying hypothesis is that such an approach will make it easier for an IR system to understand the user‟s intention behind a search. But yet
Hydrophobicity is an important parameter to characterize electrical properties of insulated materials. Therefore, it is an urgent task to develop on-line instruments to identify the hydrophobicity of insulated material’s surface conveniently, quickly and accurately. For this purpose, a novel evaluation system with imageprocessingand decision tree is proposed which is basedon embedded platform. For obtaining satisfactory results, we irst propose a mixed image segmentation method to overcome the complex conditions outside, concerning non-controlled illumination, nonstandard surfaces and unixed shooting angle. Then we adopt four new characteristic parameters to describe the image of each sample. Finally, a classiication method basedon MultiBoost decision tree is conducted which synthesizes the merits of both AdaBoost and Wagging algorithm. Results indicate the procedures can be applied in the DSP (Digital Signal Processor) platform perfectly and better results can be obtained than those did in our previous study or that of some other research.
In Bio-Medical imageprocessing domain, content- based analysis and Information retrieval of bio- images is very critical for disease diagnosis. Content-BasedImage Analysis and Information Retrieval (CBIAIR) has become a significant part of information retrieval technology. One challenge in this area is that the ever-increasing number of bio-images acquired through the digital world makes the brute force searching almost impossible. Medical Image structural objects content and object identification plays significant role for image content analysis and information retrieval. There are basically three fundamental concepts for content-based bio-imageretrieval, i.e. visual- feature extraction, multi-dimensional indexing, andretrievalsystem process. Each image has three contents such as: colour, texture and shape features. Colour and Texture both plays important image visual features used in Content-BasedImageRetrieval to improve results. In this paper, we have presented an effective imageretrievalsystem using features like texture, shape and color, called CBIAIR (Content-BasedImage Analysis and Information Retrieval). Here, we have taken three different features such as texture, color and shape. Firstly, we have developed a new texture pattern feature for pixel based feature in CBIAIR system. Subsequently, we have used semantic color feature for color based feature and the shape based feature selection is done using the existing technique. For retrieving, these features are extracted from the query imageand matched with the feature library using the feature weighted distance. After that, all feature vectors will be stored in the database using indexing procedure. Finally, the relevant images that have less matched distance than the predefined threshold value are retrieved from the image database after adapting the K-NN classifier.
Web 2.0 places users at the core of its success and empowers them to create and manage their own content. The key aspect of interaction at Web 2.0 offers the user the power of interact with the content and with other users. The information becomes fluid rather than static  and the user can shape it. This new way of interaction, where people can communicate virtually, exceeds the physical barriers and motivates online cooperation. “It’s a story about community and collaboration on a scale never seen before” . Internet is now the biggest channel of knowledge with million of Wikipedia articles and YouTube video. In several ways, such as online messaging, comments, video posting, internet allows users to be interactive and content generator. On the other hand, face-to-face communication remains important to maintain successful interactions. On this book about effective communication, Chris (1999) says that “communication is carried out face-to-face with other individuals: asking for information and offering advice”. In scenarios where security is an issue, the process of sharing information is made “one-to-one situation”, especially in the scientific world . An intelligent balance between face-to-face and web-based collaboration empowers researchers with a rich set of tools to gain valuable collaborations and expands contacts.
Brazilian rum (also known as cachaça) is the third most commonly consumed distilled alcoholic drink in the world, with approximately 2.5 billion liters produced each year. It is a traditional drink with refined features and a delicate aroma that is produced mainly in Brazil but consumed in many countries. It can be aged in various types of wood for 1–3 years, which adds aroma and a distinctive flavor with different characteristics that affect the price. A research challenge is to develop a cheap automatic recognition sys- tem that inspects the finished product for the wood type and the aging time of its production. Some clas- sical methods use chemical analysis, but this approach requires relatively expensive laboratory equipment. By contrast, the system proposed in this paper captures image signals from samples and uses an intelligent classification technique to recognize the wood type and the aging time. The classification system uses an ensemble of classifiers obtained from different wavelet decompositions. Each classifier is obtained with different wavelet transform settings. We compared the proposed approach with classical methods basedon chemical features. We analyzed 105 samples that had been aged for 3 years and we showed that the proposed solution could automatically recognize wood types and the aging time with an accuracy up to 100.00% and 85.71% respectively, and our method is also cheaper.
Digital image inpainting is an important issue in the domain of image restoration andan international interesting research topic in recent years. A lot of papers onimage inpainting were published. Although the presented algorithms to modify damaged regions of image had their merits, they only had some specific aims and at present there isn’t a universal algorithm model of “guarantee to cure all diseases” and it is not an easy matter. The chief reason for the problem is that the segmentation of image is still a difficult issue due to the segmentation algorithms are some particular question- oriented at present, and the damaged regions of image must be segmented before inpainted. Segmentation will first affect the results of inpainting. Therefore, it is more difficult to get a universal algorithm to solve all of the problems. We attempt to design a Thangka image inpainting system which can inpaint different kinds of damaged Thangka image, which is a more powerful image inpainting system than the single algorithm’s systemon the basis of previous work . This system, in which integrated various inpainting algorithms, can inpaint various kinds of
During this period, there were numerous tours of musical bands, orchestras such the concerts of cords quartet “Transilvania”, of the folk music formation “Ciocârlia” and the participation in 2003 of the Helsinki Symphonic Orchestra at the “George Enescu Festival” where compositions by the most important composers from two countries, Jan Sibelius and George Enescu, were performed. Another musical event dedicated to George Enescu took place in 2007, in Turku, at the Philharmonic Concert Hall, on the occasion of the anniversary of 125 years from the birth of the Romanian composer