View: 1-15 | 16-30 | 31-45 | 46-60 | 61-69

 

Weather Predictor for Feedforward Control Working on the Summer Data 

A. Mountis and G. Levermore

Univ. of Manchester, UK

Pages: 334-340

Abstract: For the control of heating and cooling of buildings a weather predictor with and without limited knowledge is to be developed. This project will emphasize on the summer data and thus on the control of cooling of buildings. Traditional weather forecast by simulating the atmosphere using systems of mathematical equations which is known as Numerical Weather Prediction (NWP) is not good enough to create an optimal controller for the heating and cooling of buildings. That is why a predictor adjusted to the needs for such an engineering use, is to be created. Weather in general is a completely non linear system but studying the historical data gathered from a specific place for a huge period of time, some patterns can be identified. The problem becomes –to some extent– pattern recognition and regression task, hoping that these patterns revealed in the past information will keep repeating again in the future. Weather data is available and artificial neural networks and fuzzy logic systems are selected to be used amongst the several methods for creating models able to predict.


Image De-Noising Based on the Statistical Modelling of Wavelet Coefficients and Quad-Tree Decomposition 

J. Ellinas1 and D. Manolakis2  

1Technol. Educ. Inst. of Piraeus, Greece,  2Alexander Technol. Educ. Inst. of Thessaloniki, Greece

Pages: 341-347

Abstract: This paper proposes a spatially adaptive statistical model for wavelet image coefficients in order to perform image de-noising. The wavelet coefficients are modelled as zero-mean Gaussian random variables with high local correlation. This model is developed in a Bayesian framework, where a Maximum Likelihood (ML) estimator evaluates the variance of the blocks to which the wavelet subbands have been segmented. Then, applying the Minimum Mean Squared Error (MMSE) estimation procedure, the original or de-noised wavelet image coefficients are estimated. The reliable estimation of local variance is performed by making the assumption that variance is locally smooth. The validity of this assumption is boosted by segmenting the wavelet subbands into blocks of variable size with two methods. The first method employs image quad-tree decomposition and transfers linearly the resulted tree on the wavelet subbands. This decomposition identifies object boundaries and defines more accurately the regions of smooth variance instead of dividing them in to blocks of fixed size. The second method performs quad-tree decomposition of every subband with a variance splitting criterion. The subbands are segmented into blocks of nearly constant variance, so that the transform coefficients to be approximated as i.i.d random variables. The extensive experimental evaluation shows that the proposed scheme demonstrates very good performance as far as PSNR measures and visual quality are concerned with respect to others state of the art de-noising schemes.


A Hardware Structure for Time-to-Impact Computation Using Log-Polar Images

A. Gasteratos, P. Gonidis and I. Andreadis

Democritus Univ. of Thrace, Xanthi, Greece

Pages: 348-353

Abstract: This paper describes the design of a new hardware system for the optical flow and time-to-impact computation with real-time response. This is achieved by estimating the optical flow of the object on the camera plane. In order to reduce the image size without losing significant information, the log-polar transformation is utilized. The system employs a sequence of log-polar images for the optical flow and time-to-impact computation of a moving object. The algorithm that has been implemented belongs to “differential techniques” category, which is suitable for parallel computation of the parameters. The structure of the implementation allows the processing of grey-level log-polar images of 45x60 pixels in real time (25 frames per second).


Towards Text Recognition in Natural Scene Images

B. Gatos, I. Pratikakis and S. Perantonis

Inst. of Informatics and Telecommunications, NCSR “DEMOKRITOS”, Athens, Greece

Pages: 354-359

Abstract: In this paper, we propose a novel methodology for text detection in natural scene images. The proposed methodology is based on an efficient binarization and enhancement technique followed by a suitable connected component analysis procedure. Image binarization successfully processes natural scene images having shadows, non-uniform illumination, low contrast and large signaldependent noise. Connected component analysis is used to define the final binary images that mainly consist of text regions. The proposed methodology results in increased success rates for commercial OCR engines. Experimental results based on a public database of natural scene images prove the efficiency of the proposed approach.


Image Analysis Using Moments 

L. Kotoulas and I. Andreadis

Democritus Univ. of Thrace, Xanthi, Greece

Pages: 360-364

Abstract: Moments of images provide efficient local descriptors and have been used extensively in image analysis applications. Their main advantage is their ability to provide invariant measures of shape. In this work, we present an overview of the most commonly used image moments, as well as a hardware architecture capable of fast calculation of geometric, Zernike and Tchebichef moments.


Plant Leaves Classification Based on Morphological Features and a Fuzzy Surface Selection Technique

P. Tzionas1, S. Papadakis2  and D. Manolakis1

1Alexander Technol. Educ. Inst. of Thessaloniki, Greece,  2Technol. Educ. Inst. of Kavala, Greece

Pages: 365-370

Abstract: The design and implementation of an artificial vision system that extracts specific geometrical and morphological features from plant leaves is presented in this paper. A subset of significant image features are identified using a novel feature selection approach. This approach reduces the dimensionality of the feature space leading to a simplified classification scheme appropriate for real time classification applications. A feedforward neural network is employed to perform the main classi- fication task. The proposed system exhibits size and orientation invariance with respect to the samples and it can operate successfully even with leaves samples that are deformed due to drought or due a number of holes drilled in them. A considerably high classification ratio of 99% was achieved, even for the classification of deformed leaves.


Content-Based Image Retrieval Using Cellular Automata

K. Konstantinidis, G. Sirakoulis and I. Andreadis

Democritus Univ. of Thrace, Xanthi, Greece

Pages: 371-375

Abstract: Content-based Image Retrieval (CBIR) is generally known as a collection of techniques for retrieving images on the basis of features, such as color, texture and shape. An efficient tool in CBIR is that of image histograms. In this paper a new image retrieval method is proposed with the use of histograms in conjunction with cellular automata (CAs). The main thrust of this paper is the classification of the images in the database by CAs and the retrieval of the desired images by a simple histogram extracted from the hue component of the HSV color space. Moreover, because of the CAs local rule simplicity, the VLSI implementation of the proposed CA algorithm is straightforward.


Enhancement of Sight Effectiveness by Dual Infrared System: Evaluation of Image Fusion Strategies

G. Corsini1, M. Diani1, A. Masini1 and M. Cavallini2

1Univ. di Pisa, Italy,  2Galileo Avionica, Firenze, Italy

Pages: 376-381

Abstract: The problem of objective evaluation of multisensor image fusion strategies is analysed for the design of a dual infrared system. Such a system should be used to enhance the sight effectiveness in assisting a driver or a pilot in bad visibility conditions. Two no-reference indexes are used to quantify the performance of different image fusion methods. Numerical results are presented and discussed in terms of the quality of the fused images.


User Profiling for Fraud Detection in Telecommunication Networks

C. Hilas1 and J. Sahalos2

1Technol. Educ. Inst. of Serres, Greece,  2Aristotle Univ. of Thessaloniki, Greece

Pages: 382-387

Abstract: Telecommunications fraud is increasing dramatically each year resulting in loss of a large amount of euros worldwide. A statistical machine learning method is presented that constructs user profiles for the detection of fraudulent activities in telecommunications networks. The approach presented here can be used for the detection of superimposed or hacking fraud, works well for mid-term decisions and cannot be used for on-line account comparison.

 

 

View: 1-15 | 16-30 | 31-45 | 46-60 | 61-69 | Back to top