Matching entries: 0
settings...
Author Title Year Journal/Proceedings Reftype DOI/URL
Faria, F.A., Perre, P., Zucchi, R.A., Lewinsohn, T., Rocha, A. and da S. Torres, R. Uso de Técnicas de Aprendizagem para Classificação Automática de Moscas-das-frutas (Diptera, Tephritidae) 2012 (IC-12-010)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: As moscas-das-frutas são pragas de importância quarentenária a nível mundial, dentre as quais se destacam algumas espécies de Anastrepha, que atacam um grande número de frutíferas e estão amplamente distribuídas pelo Brasil. A identifica¸ cão das espécies de moscas-das frutas é baseada nos caracteres morfológicos do mesonoto, asa e acúleo. Nos últimos anos, têm sido desenvolvidas ferramentas que complementam a taxonomia tradicional na identificação de algumas espécies de insetos. Além de diminuir o tempo gasto pelos especialistas, essas novas ferramentas podem permitir que um número maior de pesquisadores estude e/ou identifique esses insetos. Sendo assim, o presente estudo tem como objetivo testar a eficácia de novas técnicas para identificação de três espécies de moscas-das-frutas. Neste trabalho, foram aplicadas e comparadas técnicas de análise de imagens e aprendizagem de máquina na tarefa de classificação de moscas-das-frutas, visando obter dados que possam futuramente servir de base para o desenvolvimento de sistemas de identificação automática dessas espécies utilizando imagens de asas e acúleos.

BibTeX:
@techreport{Faria2012IC-12-10,
  author = {Fabio A. Faria and Paula Perre and Roberto Antonio Zucchi and Thomas Lewinsohn and Anderson Rocha and Ricardo da S. Torres},
  title = {Uso de Técnicas de Aprendizagem para Classificação Automática de Moscas-das-frutas (Diptera, Tephritidae)},
  school = {Institute of Computing, University of Campinas},
  year = {2012},
  number = {IC-12-010},
  url = {http://www.ic.unicamp.br/ reltech/2012/12-10.pdf}
}
      


Almeida, J., Pinto-Cáceres, S.M., da S. Torres, R. and Leite, N.J. Intuitive Video Browsing Along Hierarchical Trees 2011 (IC-11-06)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Recent advances in technology have increased the availability of video data, creating a strong requirement for efficient systems to manage those materials. Making efficient use of video information requires that data be accessed in a user-friendly way. Ideally, one would like to perform video search using an intuitive tool. Most of existing browsers for the interactive search of video sequences, however, have employed a too rigid layout to arrange the results, restricting users to explore the results using list- or grid-based layouts. In this paper, we present a novel approach for the interactive search that displays the result set in a flexible manner. The proposed method is based on a hierarchical structure called Divisive-Agglomerative Hierarchical Clustering (DAHC). It is able to group together videos with similar content and to organize the result set in a well-defined tree. This strategy makes the navigation more coherent and engaging to users.

BibTeX:
@techreport{Almeida2011IC-11-06,
  author = {Jurandy Almeida and Sheila M. Pinto-Cáceres and Ricardo da S. Torres and Neucimar J. Leite},
  title = {Intuitive Video Browsing Along Hierarchical Trees},
  school = {Institute of Computing, University of Campinas},
  year = {2011},
  number = {IC-11-06},
  url = {http://www.ic.unicamp.br/ reltech/2011/11-06.pdf}
}
      


Almeida, J., Leite, N.J. and da S. Torres, R. Searching in High-Dimensional Metric Spaces using BP-Trees 2011 (IC-11-08)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Similarity search in high-dimensional metric spaces is a key operation in many applications, such as multimedia databases, image retrieval, object recognition, and others. The high dimensionality of the data requires special index structures to facilitate the search. A problem regarding the creation of suitable index structures for high-dimensional data is the relationship between the geometry of the data and the organization of an index structure. Most of existing indexes are constructed by partitioning the dataset using distance-based criteria. However, those methods either produce disjoint partitions, but ignore the distribution properties of the data; or produce non-disjoint groups, which greatly affect the search performance. In this paper, we study the performance of a new index structure, called Ball-and-Plane tree (BP-tree), which overcomes the above disadvantages. BP-tree is constructed by recursively dividing the dataset into compact clusters. Different from other techniques, it integrates the advantages of both disjoint and non-disjoint paradigms in order to achieve a structure of tight and low overlapping clusters, yielding significantly improved performance. Results obtained from an extensive experimental evaluation with real-world datasets show that BP-tree consistently outperforms the state-of-the-art solutions. In addition, BP-tree scales up well, exhibiting sublinear performance with growing number of objects in the database.

BibTeX:
@techreport{Almeida2011IC-11-08,
  author = {Jurandy Almeida and Neucimar J. Leite and Ricardo da S. Torres},
  title = {Searching in High-Dimensional Metric Spaces using BP-Trees},
  school = {Institute of Computing, University of Campinas},
  year = {2011},
  number = {IC-11-08},
  url = {http://www.ic.unicamp.br/ reltech/2011/11-08.pdf}
}
      


Kozievitch, N.P., Fox, E. and da S. Torres, R. Analyzing Compound Object Technologies from the 5S Perspective 2011 (IC-11-01)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: There are many applications which need support for compound information. Thus, we need new mechanisms for managing data integration; aids for creating references, links and annotations; and services for clustering, organizing and reusing compound objects (COs) and their components. Few attempts have been made to formally characterize compound information, related services, and technologies. We propose the description and interplay of technologies for handling compound information taking advantage of the formalization proposed by the 5S Framework. This paper: (1) analyzes technologies which manage compound information (DCC, Buckets, OAI-ORE); (2) uses 5S formal definitions for describing them; (3) presents a case study, illustrating how CO technologies and the 5S Framework can fit together to support exploration of compound information.

BibTeX:
@techreport{Kozievitch2011IC-11-01,
  author = {Nádia P. Kozievitch and Edward Fox and Ricardo da S. Torres},
  title = {Analyzing Compound Object Technologies from the 5S Perspective},
  school = {Institute of Computing, University of Campinas},
  year = {2011},
  number = {IC-11-01},
  url = {http://www.ic.unicamp.br/ reltech/2011/11-01.pdf}
}
      


dos Santos, J.A., Gosselin, P.-H., Philipp-Foliguet, S., da S. Torres, R. and ao, A.X.F. Multi-Scale Classification of Remote Sensing Images 2011 (IC-11-20)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: A huge effort has been applied in image classification to create high quality thematic maps and to establish precise inventories about land cover use. The peculiarities of Remote Sensing Images (RSIs) combined with the traditional image classification challenges made RSIs classification a hard task. Our aim is to propose a kind of boost-classifier adapted to multi-scale segmentation. We use the paradigm of boosting, whose principle is to combine weak classifiers to build an efficient global one. Each weak classifier is trained for one level of the segmentation and one region descriptor. We have proposed and tested weak classifiers based on linear SVM and region distances provided by descriptors. The experiments were performed on a large image of coffee plantations. We have shown in this paper that our approach based on boosting can detect the scale and set of features best suited to a particular training set. We have also shown that hierarchical multi-scale analysis is able to reduce training time and to produce a stronger classifier. We compare the proposed methods with a baseline based on SVM with RBF kernel. The results show that the proposed methods outperform the baseline.

BibTeX:
@techreport{Santos2011IC-11-20,
  author = {Jefersson Alex dos Santos and Philippe-Henri Gosselin and Sylvie Philipp-Foliguet and Ricardo da S. Torres and Alexandre Xavier Falcão},
  title = {Multi-Scale Classification of Remote Sensing Images},
  school = {Institute of Computing, University of Campinas},
  year = {2011},
  number = {IC-11-20},
  url = {http://www.ic.unicamp.br/ reltech/2011/11-20.pdf}
}
      


Almeida, J., da S. Torres, R. and Leite, N.J. BP-tree: An Efficient Index for Similarity Search in High-Dimensional Metric Spaces 2010 (IC-10-27)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Similarity search in high-dimensional metric spaces is a key operation in many applications, such as multimedia databases, image retrieval, object recognition, and others. The high dimensionality of the data requires special index structures to facilitate the search. Most of existing indexes are constructed by partitioning the data set using distance-based criteria. However, those methods either produce disjoint partitions, but ignore the distribution properties of the data; or produce non-disjoint groups, which greatly affect the search performance. In this paper, we study the performance of a new index structure, called Ball-and-Plane tree (BP-tree), which overcomes the above disadvantages. BP-tree is constructed by recursively dividing the data set into compact clusters. Distinctive from other techniques, it integrates the advantages of both disjoint and non-disjoint paradigms in order to achieve a structure of tight and low overlapping clusters, yielding significantly improved performance. Results obtained from an extensive experimental evaluation with real-world data sets show that BP-tree consistently outperforms state-of-the-art solutions.

BibTeX:
@techreport{Almeida2010IC-10-27,
  author = {Jurandy Almeida and Ricardo da S. Torres and Neucimar J. Leite},
  title = {BP-tree: An Efficient Index for Similarity Search in High-Dimensional Metric Spaces},
  school = {Institute of Computing, University of Campinas},
  year = {2010},
  number = {IC-10-27},
  url = {http://www.ic.unicamp.br/ reltech/2010/10-27.pdf}
}
      


Kozievitch, N.P., Codio, S., Francois, J.A., Fox, E. and da S. Torres, R. Exploring CBIR concepts in the CTRnet Project 2010 (IC-10-32)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: The project Crisis, Tragedy and Recovery Network (CTRnet) is an effort to build an integrated distributed digital library for providing a rich suite of CTR-related services. This report describes an independent study conducted at Virginia Polytechnic Institute and State University, consisting of collecting and archiving information related to the Haiti earthquake, later used to explore content-based image retrieval (CBIR) concepts. The objective was to collect and categorize relevant pictures related to the earthquake, followed by the exploration of practical CBIR concepts, such as descriptors, feature vectors, and experiment design.

BibTeX:
@techreport{Kozievitch2010IC-10-32,
  author = {Nádia P. Kozievitch and Sherley Codio and Jennifer A. Francois and Edward Fox and Ricardo da S. Torres},
  title = {Exploring CBIR concepts in the CTRnet Project},
  school = {Institute of Computing, University of Campinas},
  year = {2010},
  number = {IC-10-32},
  url = {http://www.ic.unicamp.br/ reltech/2010/10-32.pdf}
}
      


Li, L.T. and da Silva Torres, R. Coping with geographical relationships in Web searches 2010 (IC-10-04)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: This work describes ongoing research which aims to evaluate the limitations of existing tools to support geographic information retrieval on the Web. More specifically, we are interested in assessing how well queries involving geographic relationships can be specified and executed using existing Web tools. Our evaluation considered three different aspects: the relevance of the results and the time and complexity of the steps necessary to obtain these results. More than thirty Web users performed five tasks, considering five scenarios. Experiment results demonstrate that existing Web tools are not enough integrated to meet user needs regarding the specification and execution of queries involving spatial relationships.

BibTeX:
@techreport{Lin2010IC-10-04,
  author = {Lin Tzy Li and Ricardo da Silva Torres},
  title = {Coping with geographical relationships in Web searches},
  school = {Institute of Computing, University of Campinas},
  year = {2010},
  number = {IC-10-04},
  url = {http://www.ic.unicamp.br/ reltech/2010/10-04.pdf}
}
      


Almeida, J., Minetto, R., Almeida, T.A., da S. Torres, R. and Leite, N.J. Robust Estimation of Camera Motion using Local Invariant Features 2009 (IC-09-12)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Most of existing techniques to estimate camera motion are based on analysis of the optical flow. However, the estimation of the optical flow supports only a limited amount of scene motion. In this report, we present a novel approach to estimate camera motion based on analysis of local invariant features. Such features are robust across a substantial range of affine distortion. Experiments on synthesized video clips with a fully controlled environment show that our technique is more effective than the optical flow-based approaches for estimating camera motion with a large amount of scene motion.

BibTeX:
@techreport{Almeida2009IC-09-12,
  author = {Jurandy Almeida and Rodrigo Minetto and Tiago A. Almeida and Ricardo da S. Torres and Neucimar J. Leite},
  title = {Robust Estimation of Camera Motion using Local Invariant Features},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-12},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-12.pdf}
}
      


Almeida, J., Minetto, R., Almeida, T.A., da S. Torres, R. and Leite, N.J. Robust Estimation of Camera Motion using Optical Flow Models 2009 (IC-09-24)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: The estimation of camera motion is one of the most important aspects for video processing, analysis, indexing, and retrieval. Most of existing techniques to estimate camera motion are based on optical flow methods in the uncompressed domain. However, to decode and to analyze a video sequence is extremely time-consuming. Since video data are usually available in MPEG-compressed form, it is desirable to directly process video material without decoding. In this paper, we present a novel approach for estimating camera motion in MPEG video sequences. Our technique relies on linear combinations of optical flow models. The proposed method first creates prototypes of optical flow, and then performs a linear decomposition on the MPEG motion vectors, which is used to estimate the camera parameters. Experiments on synthesized and real-world video clips show that our technique is more effective than the state-of-the-art approaches for estimating camera motion in MPEG video sequences.

BibTeX:
@techreport{Almeida2009IC-09-24,
  author = {Jurandy Almeida and Rodrigo Minetto and Tiago A. Almeida and Ricardo da S. Torres and Neucimar J. Leite},
  title = {Robust Estimation of Camera Motion using Optical Flow Models},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-24},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-24.pdf}
}
      


Faria, F.A., Veloso, A., de Almeida, H.M., Valle, E., da S. Torres, R., Gonçalves, M.A. and Jr., W.M. Learning to Rank for Content-Based Image Retrieval 2009 (IC-09-37)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: In Content-based Image Retrieval (CBIR), accurately ranking the returned images is of paramount importance, since it is common-sense that users consider mostly the topmost results. The typical ranking strategy used by many CBIR systems is to employ image content descriptors, so that returned images that are most similar to the query image are placed higher in the rank. While this strategy is well accepted and widely used, improved results may be obtained by combining multiple image descriptors. In this paper we explore this idea, and introduce algorithms that learn to combine information coming from different descriptors. The proposed learning to rank algorithms are based on three diverse learning techniques: Support Vector Machines (CBIR-SVM), Genetic Programming (CBIR-GP), and Association Rules (CBIR-AR). Eighteen image content descriptors (color, texture, and shape information) are used as input and provided as training to the learning algorithms. We performed a systematic evaluation involving two complex and heterogeneous image databases (Corel e Caltech) and two evaluation measures (Precision and MAP). The empirical results show that all learning algorithms provide significant gains when compared to the typical ranking strategy in which descriptors are used in isolation. We concluded that, in general, CBIR-AR and CBIR-GP outperforms CBIR-SVM. A fine-grained analysis revealed the lack of correlation between the results provided by CBIR-AR and the results provided by the other two algorithms, which indicates the opportunity of an advantageous hybrid approach.

BibTeX:
@techreport{Faria2009IC-09-37,
  author = {Fabio A. Faria and Adriano Veloso and Humberto Mossri de Almeida and Eduardo Valle and Ricardo da S. Torres and Marcos A. Gonçalves and Wagner Meira Jr.},
  title = {Learning to Rank for Content-Based Image Retrieval},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-37},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-37.pdf}
}
      


Kozievitch, N.P., da S. Torres, R., Falcão, T., Ramos, E., Andrade, F., Allegretti, S.M., Ueta, M.T., Madi, R.R., Murthy, U., Fox, E.A., Chen, Y. and Hallerman, E. Evaluation of a Tablet PC Image Annotation and Retrieval Tool in the Parasitology Domain 2009 (IC-09-23)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: The project Deployment and Assessment of an Image Annotation and Retrieval Tool has the objective of specifying and implementing an application for image support annotation and search (based on a textual and a visual description) in the biodiversity domain. This technical report presents the activities related to the use of the tablet PC tool in the parasitology domain at Unicamp. The objective of this tool is to help the comparison of morphological characteristics among different species. The report is divided into activities accomplished, application setup and specific features, followed by experimental results and conclusion. Preliminary results showed that students regarded the tool as being very useful, contributing as an alternative learning approach.

BibTeX:
@techreport{Kozievitch2009IC-09-23,
  author = {Nádia. P. Kozievitch and Ricardo da S. Torres and Tiago Falcão and Evandro Ramos and Felipe Andrade and Silmara M. Allegretti and Marlene T. Ueta and Rubens R. Madi and Uma Murthy and Edward A. Fox and Yinlin Chen and Eric Hallerman},
  title = {Evaluation of a Tablet PC Image Annotation and Retrieval Tool in the Parasitology Domain},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-23},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-23.pdf}
}
      


Li, L.T. and da Silva Torres, R. Revisitando os Desafios da Recuperação de Informação Geográfica na Web 2009 (IC-09-18)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: The geographic information is part of people's daily life. There is a huge amount of information on the Web about or related to geographic entities and people are interested in localizing them on maps. Nevertheless, the conventional Web search engines, which are keywords-driven mechanisms, do not support queries involving spatial relationships between geographic entities. This paper revises the Geographic Information Retrieval (GIR) area and restates its research challenges and opportunities, based on a proposed architecture for executing Web queries involving spatial relationships and an initial implementation of that.

BibTeX:
@techreport{Li2009IC-09-18,
  author = {Lin Tzy Li and Ricardo da Silva Torres},
  title = {Revisitando os Desafios da Recuperação de Informação Geográfica na Web},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-18},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-18.pdf}
}
      


dos Santos, J.A., Penatti, O.A.B. and da S. Torres, R. Evaluating the Potential of Texture and Color Descriptors for Remote Sensing Image Retrieval and Classification 2009 (IC-09-47)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Classifying Remote Sensing Images (RSI) is a hard task. There are automatic approaches whose results normally need to be revised. The identification and polygon extraction tasks usually rely on applying classification strategies that exploit visual aspects related to spectral and texture patterns identified in RSI regions. There are a lot of image descriptors proposed in the literature for content-based image retrieval purposes that can be useful for RSI classification. This paper presents a comparative study to evaluate the potential of using successful color and texture image descriptors for remote sensing retrieval and classification. Seven descriptors that encode texture information and twelve color descriptors that can be used to encode spectral information were selected. We highlight the main characteristics and perform experiments to evaluate the effectiveness of these descriptors. To evaluate descriptors in classification tasks, we also proposed a methodology based on KNN classifier. Experiments demonstrate that Joint Auto-Correlogram (JAC), Color Bitmap, Invariant Steerable Pyramid Decomposition (SID) and Quantized Compound Change Histogram (QCCH) yield the best results

BibTeX:
@techreport{Santos2009IC-09-47,
  author = {Jefersson A. dos Santos and Otávio A. B. Penatti and Ricardo da S. Torres},
  title = {Evaluating the Potential of Texture and Color Descriptors for Remote Sensing Image Retrieval and Classification},
  school = {Institute of Computing, University of Campinas},
  year = {2009},
  number = {IC-09-47},
  url = {http://www.ic.unicamp.br/ reltech/2009/09-47.pdf}
}
      


dos Santos, J.A., Ferreira, C.D. and da S. Torres, R. A Genetic Programming Approach for Relevance Feedback in Region-based Image Retrieval Systems 2008 (IC-08-19)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: This paper presents a new relevance feedback method for content-based image retrieval using local image features. This method adopts a genetic programming approach to learn user preferences and combine the region similarity values in a query session. Experiments demonstrate that the proposed method yields more effective results than the Local Aggregation Pattern (LAP)-based relevance feedback technique.

BibTeX:
@techreport{Santos2008IC-08-19,
  author = {Jefferson A. dos Santos and Cristiano D. Ferreira and Ricardo da S. Torres},
  title = {A Genetic Programming Approach for Relevance Feedback in Region-based Image Retrieval Systems},
  school = {Institute of Computing, University of Campinas},
  year = {2008},
  number = {IC-08-19},
  url = {http://www.ic.unicamp.br/ reltech/2008/08-19.pdf}
}
      


Almeida, J., Rocha, A., da S. Torres, R. and Goldestein, S. Image Retrieval based on Color and Scale representative Image Regions (CSIR) 2007 (IC-07-28)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Content-based image retrieval (CBIR) is a challenging task. Common techniques use only low-level features. However, these solutions can lead to the so-called `semantic gap' problem: images with high feature similarities may be different in terms of user perception. In this paper, our objective is to retrieve images based on color cues which may present some affine transformations. For that, we present CSIR: a new method for comparing images based on discrete distributions of distinctive color and scale image regions. We validate the technique using images with a large range of viewpoints, partial occlusion, changes in illumination, and various domains.

BibTeX:
@techreport{Almeida2007IC-07-28,
  author = {Jurandy Almeida and Anderson Rocha and Ricardo da S. Torres and Siome Goldestein},
  title = {Image Retrieval based on Color and Scale representative Image Regions (CSIR)},
  school = {Institute of Computing, University of Campinas},
  year = {2007},
  number = {IC-07-28},
  url = {http://www.ic.unicamp.br/ reltech/2007/07-28.pdf}
}
      


Ferreira, C.D. and da S. Torres, R. Image Retrieval with Relevance Feedback based on Genetic Programming 2007 (IC-07-05)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: In the last years, large digital image collections are generated, manipulated, and stored in databases. In this scenery, it is very important to develop mechanisms to provide automatic means to retrieve images in an efficient and effective way. However, the subjectivity of the user perception of an image usually hampers a fully automatic search and retrieval. Relevance Feedback is one of the commonest approaches to overcome this difficult.

In this paper, a new content-based image retrieval framework with relevance feedback is proposed. This framework uses Genetic Programming (GP) to learn the user needs. The objective of this learning method is to find a function that combines different values of similarity, from distinct descriptors, and best encodes the user perception of image similarity. Several experiments are performed to validate the proposed method, aiming to compare our work with other relevance feedback techniques. The experiment results show that the proposed method outperforms all of them.



BibTeX:
@techreport{Ferreira2007IC-07-05,
  author = {Cristiano D. Ferreira and Ricardo da S. Torres},
  title = {Image Retrieval with Relevance Feedback based on Genetic Programming},
  school = {Institute of Computing, University of Campinas},
  year = {2007},
  number = {IC-07-05},
  url = {http://www.ic.unicamp.br/ reltech/2007/07-05.pdf}
}
      


Kim, S., Fox, E.A., Fan, W., North, C., Tatar, D. and da S. Torres, R. Design and Evaluation of Techniques to Utilize Implicit Rating Data in Complex Information Systems 2007 (TR-07-20)School: Computer Science Department, Virginia Tech  techreport URL 
Abstract: Research in personalization, including recommender systems, focuses on applications such as in online shopping malls and simple information systems. These systems consider user profile and item information obtained from data explicitly entered by users - where it is possible to classify items involved and to make personalization based on a direct mapping from user or user group to item or item group. However, in complex, dynamic, and professional information systems, such as Digital Libraries, additional capabilities are needed to achieve personalization to support their distinctive features: large numbers of digital objects, dynamic updates, sparse rating data, biased rating data on specific items, and challenges in getting explicit rating data from users. In this report, we present techniques for collecting, storing, processing, and utilizing implicit rating data of Digital Libraries for analysis and decision support. We present our pilot study to find virtual user groups using implicit rating data. We demonstrate the effectiveness of implicit rating data for characterizing users and finding virtual user communities, through statistical hypothesis testing. Further, we describe a visual data mining tool named VUDM (Visual User model Data Mining tool) that utilizes implicit rating data. We provide the results of formative evaluation of VUDM and discuss the problems raised and plans for further studies.

BibTeX:
@techreport{Kim2007VT-TR-07-20,
  author = {Seonho Kim and Edward A. Fox and Weiguo Fan and Chris North and Deborah Tatar and Ricardo da S. Torres},
  title = {Design and Evaluation of Techniques to Utilize Implicit Rating Data in Complex Information Systems},
  school = {Computer Science Department, Virginia Tech},
  year = {2007},
  number = {TR-07-20},
  url = {http://eprints.cs.vt.edu/archive/00000980/}
}
      


Rocha, A., Almeida, J., da S. Torres, R. and Goldestein, S. A New Hybrid Clustering Approach for Image Retrieval 2007 (IC-07-29)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: In this paper, we present a new Hybrid Hierarchical Clustering approach for Image Retrieval. Our method combines features from both divisive and agglomerative clustering paradigms in order to yield good-quality clustering solutions with reduced computational cost. We provide several experiments showing that our technique reduces the number of required comparisons to perform a retrieval without significant loss in effectiveness when compared to flat-based solutions.

BibTeX:
@techreport{Rocha2007IC-07-29,
  author = {Anderson Rocha and Jurandy Almeida and Ricardo da S. Torres and Siome Goldestein},
  title = {A New Hybrid Clustering Approach for Image Retrieval},
  school = {Institute of Computing, University of Campinas},
  year = {2007},
  number = {IC-07-29},
  url = {http://www.ic.unicamp.br/ reltech/2007/07-29.pdf}
}
      


Montoya-Zegarra, J.A., Leite, N.J. and da S. Torres, R. Wavelet-based Feature Extraction for Fingerprint Image Retrieval 2006 (IC-06-12)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: This paper presents a novel approach to fingerprint retrieval for personal identification by joining three image retrieval tasks, namely, feature extraction, similarity measurement, and feature indexing, into a wavelet-based fingerprint retrieval system.

We propose the use of different types of Wavelets for representing and describing the textural information present in fingerprint images. For that purposes, the feature vectors used to characterize the fingerprints are obtained by computing the mean and the standard deviation of the decomposed images in the Wavelet domain. These feature vectors are used to retrieve the most similar fingerprints given a query image, while their indexation is used to reduce the search spaces of image candidates. The different types of Wavelets used in our study include: Gabor Wavelets (GWs), Tree-Structured Wavelet Decomposition using both Orthogonal Filter Banks (TOWT) and Bi-orthogonal Filter Banks (TBOWT), as well as the Steerable Wavelets.

To evaluate the retrieval accuracy of the proposed approach, a total number of eight different data sets were used. Experiments also evaluated different combinations of Wavelets with six similarity measures. The results show that the Gabor Wavelets combined with the Square Chord similarity measure achieves the best retrieval effectiveness.



BibTeX:
@techreport{Montoya-Zegarra2006IC-06-12,
  author = {Javier A. Montoya-Zegarra and Neucimar J. Leite and Ricardo da S. Torres},
  title = {Wavelet-based Feature Extraction for Fingerprint Image Retrieval},
  school = {Institute of Computing, University of Campinas},
  year = {2006},
  number = {IC-06-12},
  url = {http://www.ic.unicamp.br/ reltech/2006/06-12.pdf}
}
      


da S. Torres, R., Falcão, A.X., Gonçalves, M.A., Zhang, B., Fan, W. and Fox, E.A. A New Framework to Combine Descriptors for Content-based Image Retrieval 2005 (IC-05-21)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Methods that combine image database descriptors have strong influence on the effectiveness of content-based image retrieval (CBIR) systems. Although there are many combination functions described in the image processing literature, empirical evaluation studies have shown that those functions do not perform consistently well across different contexts (queries, image collections, users). Moreover, it is often very difficult for human beings to identify optimal combination functions for a particular application. In this paper, we propose a novel framework using Genetic Programming to combine image database descriptors for CBIR. Our framework is validated through several experiments involving two image databases and a specific domain, where the images are retrieved based on the shape of their objects.

BibTeX:
@techreport{Torres2005IC-05-21,
  author = {Ricardo da S. Torres and Alexandre X. Falcão and Marcos A. Gonçalves and Baoping Zhang and Weiguo Fan and Edward A. Fox},
  title = {A New Framework to Combine Descriptors for Content-based Image Retrieval},
  school = {Institute of Computing, University of Campinas},
  year = {2005},
  number = {IC-05-21},
  url = {http://www.ic.unicamp.br/ reltech/2005/05-21.pdf}
}
      


da S. Torres, R. and Falcão, A.X. Contour Salience Descriptors for Effective Image Retrieval and Analysis 2004 (IC-04-11)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: This work exploits the resemblance between content-based image retrieval and image analysis with respect to the design of image descriptors and their effectiveness. In this context, two shape descriptors are proposed: contour saliences and segment saliences. Contour saliences revisits its original definition, where the location of concave points was a problem, and provides a robust approach to incorporate concave saliences. Segment saliences introduces salience values for contour segments, making it possible to use an optimal matching algorithm as distance function. The proposed descriptors are compared with convex contour saliences, curvature scale space, and beam angle statistics using a fish database with 11,000 images organized in 1,100 distinct classes. The results indicate segment saliences as the most effective descriptor for this particular application and confirm the improvement of the contour salience descriptor in comparison with convex contour saliences.

BibTeX:
@techreport{Torres2004IC-04-11,
  author = {Ricardo da S. Torres and Alexandre X. Falcão},
  title = {Contour Salience Descriptors for Effective Image Retrieval and Analysis},
  school = {Institute of Computing, University of Campinas},
  year = {2004},
  number = {IC-04-11},
  url = {http://www.ic.unicamp.br/ reltech/2004/04-11.pdf}
}
      


da S. Torres, R., Medeiros, C.B., Hallerman, E.M., Gonçalves, M.A. and Fox, E.A. Integrating Image and Spatial Data for Biodiversity Information Management 2004 (IC-04-12)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: Biologists gather many kinds of data for biodiversity studies; these data are managed by distinct types of information systems. GIS-based biodiversity systems support sophisticated spatial correlations on living beings and their habitats, and spatio-temporal ecosystem modeling. Image information systems allow content-based image retrieval, to help species identification based on similarity (e.g., shape and color characteristics). Different kinds of rule-based systems support species characterization. Unfortunately, these systems (and the underlying data) are independent of each other. This paper presents a solution that seamlessly combines these functionalities, supporting queries that merge textual descriptions, spatial correlations and content-based predicates. The solution is being implemented at Virginia Tech, for identification and data retrieval, supporting management of fish species. It takes advantage of innovations in Digital Library technology to combine networked collections of heterogeneous data under integrated management.

BibTeX:
@techreport{Torres2004IC-04-12,
  author = {Ricardo da S. Torres and Claudia B. Medeiros and Eric M. Hallerman and Marcos A. Gonçalves and Edward A. Fox},
  title = {Integrating Image and Spatial Data for Biodiversity Information Management},
  school = {Institute of Computing, University of Campinas},
  year = {2004},
  number = {IC-04-12},
  url = {http://www.ic.unicamp.br/ reltech/2004/04-12.pdf}
}
      


da S. Torres, R., Medeiros, C.B., Gonçalves, M.A. and Fox, E.A. An OAI-based Digital Library Framework for Biodiversity Information Systems 2004 (TR-04-01)School: Computer Science Department, Virginia Tech  techreport URL 
Abstract: Biodiversity information systems (BISs) involve all kinds of heterogeneous data, which include ecological and geographical features. However, available information systems offer very limited support for managing such data in an integrated fashion, and integration is often based on geographic coordinates alone. Furthermore, such systems do not fully support image content management (e.g., photos of landscapes or living organisms), a requirement of many BIS end-users. In order to meet their needs, these users - e.g., biologists, environmental experts - often have to alternate between distinct biodiversity and image information systems to combine information extracted from them. This cumbersome operational procedure is forced on users by lack of interoperability among these systems. This hampers the addition of new data sources, as well as cooperation among scientists. The approach provided in this paper to meet these issues is based on taking advantage of advances in Digital Library (DL) innovations to integrate networked collections of heterogeneous data. It focuses on creating the basis for a biodiversity information system under the digital library perspective, combining new techniques of content-based image retrieval and database query processing mechanisms. This approach solves the problem of system switching, and provides users with a flexible platform from which to tailor a BIS to their needs.

BibTeX:
@techreport{Torres2004VT-TR-04-01,
  author = {Ricardo da S. Torres and Claudia B. Medeiros and Marcos A. Gonçalves and Edward A. Fox},
  title = {An OAI-based Digital Library Framework for Biodiversity Information Systems},
  school = {Computer Science Department, Virginia Tech},
  year = {2004},
  number = {TR-04-01},
  url = {http://eprints.cs.vt.edu/archive/00000680/}
}
      


da S. Torres, R., Falcão, A.X. and da F. Costa, L. A Graph-based Approach for Multiscale Shape Analysis 2003 (IC-03-03)School: Institute of Computing, University of Campinas  techreport URL 
Abstract: This paper presents the advantages of computing two recently proposed shape descriptors, multiscale fractal dimension and contour saliences, using the image foresting transform--a graph-based approach to the design of image processing operators. It introduces a robust approach to estimate contour saliences (peaks of high curvature) by exploiting the relation between contour and skeleton. The paper also compares both shape descriptors to fractal dimension, Fourier descriptors, and moment invariants with respect to their invariance to object characteristics that belong to a same class (compact-ability) and to their discriminatory ability to separate objects that belong to distinct classes (separability).

BibTeX:
@techreport{Torres2003IC-03-03,
  author = {Ricardo da S. Torres and Alexandre X. Falcão and Luciano da F. Costa},
  title = {A Graph-based Approach for Multiscale Shape Analysis},
  school = {Institute of Computing, University of Campinas},
  year = {2003},
  number = {IC-03-03},
  url = {http://www.ic.unicamp.br/ reltech/2003/03-03.pdf}
}