Mihaela Gordan

With the increasing sizes of high resolution images, their storage and processing directly in the compressed domain has significantly gained importance. Algorithms for compressed domain image processing provide a powerful computational alternative to classical (pixel level) based implementations. While linear algorithms can be applied straightforward to the JPEG compressed images, this is not the case for nonlinear image processing, as

ABSTRACT This paper deals with the calibration of a stereo configuration using two different approaches: a classic perspective projection model and a novel model using fuzzy systems. The two models are mathematically formulated, are calibrated using synthetic data and eventually are compared according to several criteria. The advantages and disadvantages of each method are analyzed with respect to the possible applications of the stereo system. Therefore, this paper highlights some arguments for the choice of the appropriate calibration method of a stereo system of cameras used for non-intrusive distance measurements taking into account the final application requirements.

In today's digital circuit simulators, the voltages during the transition time are only qualitatively described. We propose here a fuzzy logic model for the TTL inverter which quantitatively describes both its static and dynamic voltage behavior, according to its double slope voltage transfer characteristic. This model can also be viewed as a basic for developing similar models for the other elementary gates

Many median filters are developed for images affected by color impulse noise. A particular approach aims to preserve fine details by noise detection followed by filtering. The color noise detection algorithms vary as principle and performance. This paper proposes a new color image filtering method from this class, which jointly applies two methods of modified fuzzy c-means clustering for the detection of noisy pixels and afterwards performs a color noise filtering on the detected pixels only. The approach shows a good noise detection performance (in terms of false acceptance and false rejection rates), and the filtering performance in terms of PSNR and details preservation is superior to other filters (including vector median filter).

Present energy problems involve issues of technology selection, placement, variation of energy services demand by location and time-of-use and new sourcing options especially renewable energy. Beside these issues, there are also, major challenges for current energy production where the cost and environmental impacts are some of the important aspects which determine the producers to consider alternative solutions. This paper describes an informational system for designing energy systems based on renewable energy (wind energy) in isolated areas. The proposed method, which refers to model, simulation and configuration of the energy system, is based on CAD (Computer Aided Design) and DDS (Decision Support System) techniques.

Underwater images analysis is a difficult task due to their specific attributes: weak and variable lighting, low contrast, blurring. Therefore powerful image analysis algorithms, application specific, must be employed to obtain good results. In this paper we propose such a novel architecture based on a support vector machine (SVM) classifier, dedicated to large underwater scenes analysis for the specific task

Ophthalmology is a significant branch of the biomedical field which requires computer-aided automated techniques for pathology identification. Within this framework, an important concern is the accurate segmentation of the retinal blood vessels. A reference approach in the literature to this task consists in the classification of the pixels as vessels or non-vessels, using as discriminative features the green channel intensity, two-dimensional Gabor wavelet responses and some variants of LBP descriptors. However the discriminative power of this feature set is not always sufficient to provide a really highly accurate segmentation. In this paper we propose a new approach, combining powerful machine learning classifiers: support vector machines and neural networks over the same feature set, to improve the classification accuracy by a weighted decision fusion. The experimental results obtained on the DRIVE database show that the segmentation accuracy is increased up to 94%, which is superior to similar segmentation methods from the literature using neural networks, Bayesian, unsupervised classifiers and even support vector machines individually. When these results are further combined with the output of matched filters applied on the retinal images, the segmentation accuracy is further increased, by a better identification of the fine vessels.

Speech recognition based on visual information is an emerging research field. We propose here a new system for the recognition of visual speech based on support vector machines which proved to be powerful classifiers in other visual tasks. We use support vector machines to recognize the mouth shape corresponding to different phones produced. To model the temporal character of the speech we employ the Viterbi decoding in a network of support vector machines. The recognition rate obtained is higher than those reported earlier when the same features were used. The proposed solution offers the advantage of an easy generalization to large vocabulary recognition tasks due to the use of viseme models, as opposed to entire word models.

The monitoring and behavioral prediction of the hydrodams and hydrotechnical sites relies on the analysis of some objective information, the large number of sensors and examination modalities renders the human inspection of this information very difficult if not even impossible. The main objective of system is to provide a solution to overcome this problem through the development of a decision support system with an expert system component, able to minimize the subjectivity of the human expert in monitoring and behavioral prediction of the hydrotechnical structures and sites. The present article describe an integrated system, decisional support based on multisensorial information fusion provided by supervisor sensors from dams and hydropower plants, related to meteorological and geophysical factors, that can achieved behavior surveillance and prediction of dams. So, the system can provide the detection, diagnosis and monitoring of a structure defects or other functional anomalies, can also, forewarn, through the prediction component, the development of danger.

Hydro-dams safety represents an important concern since their failure could be critical for the society. A key part of the hydro-dams surveillance programs is their visual inspection. However few computer vision support tools for implementing semi-automatically and objectively the visual surveillance and observation of the hydro-dams components exist. One of the issues addressed during the visual inspection, important in the preservation of a good condition of the concrete, is the examination of surface deterioration in respect to small patterned cracks and roughness on the downstream wall. This is particularly a task where digital image enhancement and analysis can bring significant benefit, not only by presenting the user with a more relevant image of the surface deterioration, but also by providing - through suitable numerical descriptors, correlated with linguistic descriptors- subjective and examiner-independent information on the surface state. The correlation of extracted numerical descriptors used to quantify the surface roughness with linguistic qualifiers of the deterioration state of the hydro-dam wall should be determined using information gathered from observers, since it must be compliant to the human expert interpretation of visual data in assessing the concrete surface deterioration. Such an approach would result in a computer vision decision support tool embedding expert knowledge, as designed, implemented and proposed in this paper. The resulting software system was verified on a set of images acquired from a Romanian hydro-dam. The compliance of the linguistic results with the human observation proves its functionality as a semi-automatic tool for hydro-dams surveillance.

Medical images visualization and interaction is critical for assisted diagnosis and treatment. There are several 3D medical image visualization and interaction frameworks, but no standard tools exist for visual interaction. These tools must gain the user's acceptance, be easy to manipulate and allow a high positioning precision during the interaction with the virtual medical volume. Among them, the virtual probes/pens gained a good acceptance. This paper presents such an interaction tool and its applications to the editing of an original or segmented volume. Its main novelty is the use of fuzzy logic in the 3D virtual probe positioning, with the advantages of a low memory usage, real-time operation and low positioning errors as compared to classical solutions. The proposed fuzzy 3D visual interaction framework integrates several functionalities to medical volume editing and visualization: 3D measurements; positioning arbitrary cutting planes; cropping a volume of interest; surface/volume smoothing; 3D morphological operations; anaglyph stereo visualization. Having a flexible architecture, many other types of processing may be easily integrated in the future.

A&QT-R 2002 (THETA 13) 2002 IEEE-TTTC International Conference on Automation, Quality and Testing, Robotics May 23-25, 2002, Cluj-Napoca, Romania ... A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION

Median filters are widely used for the reduction of impulse noise in digital images. Since particular problems appear for color images affected by color impulse noise (as color altering or imperfect noise elimination), many color median filtering methods are still developed. Among these, fuzzy median filters are reported to be highly efficient. In this paper, we propose a novel adaptive fuzzy logic-based color filtering algorithm, designed for the hue-saturation-value (HSV) space, in the form of an adaptive cascade of fuzzy logic systems. The performance of the proposed filter is superior to other fuzzy approaches, as shown by the experimental results.

Water resource management evaluation and planning is an important issue in the context of natural resources management, as it impacts the society, environment, ecology and economy. In the context of the information society, the development of comprehensive decision support components, with the capability of displaying the results of data processing in a meaningful form for the user, becomes increasingly important. This paper proposes a decision support component for water resource management assessment tasks. As many current state of the art approaches, the proposed component employs fuzzy logic in the management evaluation part, using a hierarchical process analysis strategy with qualitative reasoning. Unlike previous existing works, we aim to enhance the presentation of the assessment results by displaying them not only in the form of linguistic qualifiers assigned to the management policy, but also in a graphical form consisting in a geotypical textured map of the region, where the natural texture changes according to the evaluation result for a specific category and according to the qualifier assigned to the management policy (varying from worst to very good). This allows the user to get multiple clues on the results of the water resource management evaluation, hopefully in a more meaningful way than from the numerical/linguistic assessment alone. The proposed component was implemented for the water resource management evaluation in the Somes river basin in Romania.

Support vector machines (SVMs) are powerful classifiers, with very good recognition rates in image analysis tasks. However their computational time in the object recognition phase is often large due to the number of classifications per scene and to the feature vector size, especially when the feature space is formed from raw image data. Several methods are reported in the literature to make the classification faster, as selecting only the most significant support vectors or reducing the feature vector length by image transforms (wavelets, PCA) prior to SVM training and classification. The method we propose is different in principle. Instead of applying the transform prior to training and thus changing the representation space, we only perform a unitary orthogonal real transform in the classification phase on the resulting support vectors and on the pattern to be classified. As the inverse matrices of these transforms are exactly the transposed of the transform matrices, we mathematically prove that the dot product of any two vectors has the same expression in the original and the transformed space. This, combined with the energy compaction property of a suitable transform, leads to a faster computation of the dot products, if the transform has a fast implementation algorithm. We use the discrete cosine transform (DCT) due to its good energy compaction on digital images. Our first experiments on a face recognition application are promising: at the same recognition rate, our algorithm leads to an average 30% reduction in the number of elementary operations per classification

Efficient hydro plants represent one of the most important social structure, wielding a great impact against inhabitants. Surveillance and monitoring of hydro dams represent a serious environmental and technical problem. Limited access to some areas of hydro plants makes their inspection difficult. A recent solution is the use of visual inspection with underwater robot vehicles for the surveillance of the

This paper proposes a classification scheme of prostate cancer patients based on support vector machines (SVM) classifiers that allow including the diagnosed prostate cancer patients into risk classes, before performing radical prostatectomy, according to their medical parameters. Our objective is to assess the use of SVM in order to predict the individual result of radical prostatectomy performed on prostate cancer patients. In medicine, the balance now leans over towards practical experience, as there are more and more information and knowledge on which physicians base their decisions. The treatment options may be different from patient to patient. The surgical decision about prostate cancer is often a complex matter; thus the proposed schema is a very useful tool that allows the physician to benefit from information regarding the outcome of previous cases.