Earticle

In this paper, we propose an algorithm to locate an object with unknown coordinates based on the positive semi-definite programming in the wireless sensor networks, assuming that the squared error of the measured distance follows Gaussian distribution. We first obtain the estimator of the object location based on the maximum likelihood criterion; then considering that the estimator is a non-convex function with respect to the measured distances between the object and the anchors with known coordinates, we transform the non-convex optimization to convex one by the positive semi-definite relaxation; and finally we take the optimal solution of the convex optimization as the estimated value of the object location. Simulations results show that our algorithm is superior to the R-LS algorithm regardless of whether the object is located within the convex hull composed of the anchors.

Computer-aided diagnosis (CADx) is usually used as a second opinion by the radiologists. Improving CADx increases the treatment options and a cure is more likely. The main objective of this research is to introduce a new method for feature extraction in order to build a CADx model to discriminate between cancers, benign, and healthy parenchyma. For feature extraction, we use both human features, which are obtained by Digital Database for Screening Mammography (DDSM), and computational features, and we propose a new feature extraction method called Square Centroid Lines Gray Level Distribution Method (SCLGM). The experimental results are obtained from a data set of 410 images taken from DDSM for different types. Our method select 31 features from 145 extracted features; 18 of the selected features are from our proposed feature extraction method (SCLGM). We used both Receiver Operating Characteristics (ROC) and confusing matrix to measure the performance. In training stage, our proposed method achieved an overall classification accuracy of 96.3%, with 92.9% sensitivity and 94.3% specificity. In testing stage, our proposed method achieved an overall classification accuracy of 89%, with 88.6% sensitivity and 83.3% specificity.

Data encryption is a vital application to ensure security in open networks like internet. Image is widely used for various purposes and one of the most popular formats of data on the web. Different image encryption algorithms are proposed so far to generate encrypted image so that it is so difficult make prediction of pixels value by attackers. This paper proposes a new framework for image encryption, a layer based method combining some most efficient image encryption algorithms. It is tried to take all encryption concerns into consider, achieve highest possible level of security while cost is already acceptable. The proposed method provides different security level to blocks of varied significance in image to consume less computational resources. Various analysis and experiments such as histogram, correlation coefficient, entropy, and computational time revealed that significant promotion in security has been achieved without compromising on the computational time.

Regional Atrophy analysis of structural magnetic resonance image (MRI) of the brain may provide quantitative evidence of different neurodegenerative diseases. This paper proposes an approach for early detection of Alzheimer disease (AD) by locating the atrophy in the brain. The paper proposes an automated computer aided system to for differential diagnosis of different neurodegenerative diseases by regional atrophy analysis such as the hippocampus that is well known to be affected in early Alzheimer disease (AD). The paper proposes in calculated factors to be used in differential diagnosis of different neurodegenerative diseases. The proposed system with its modules; preprocessing, segmentation, regional analysis, detection and evaluation yielded promising results.

This paper presents a statistical learning algorithm based on Support Vector Machines (SVMs) for the classification of Malayalam Consonant – Vowel (CV) speech unit in noisy environments. We extend SVM for multiclass classification using Decision Directed Acyclic Graph Support Vector Machine (DDAGSVM) algorithm. For classification, acoustical features are extracted using Wavelet Transform (WT) based Normalized Wavelet Hybrid Features (NWHF) by combining both Classical Wavelet Decomposition (CWD) and Wavelet Packet Decomposition (WPD) along with z – score normalization. An optimum mother wavelet for the present speech database is selected as db2 by trial and error approach. The classification results are then compared with both Artificial Neural Networks (ANNs) and k – Nearest Neighborhood (k – NN) classifiers. The results indicate that the DDAGSVM algorithm perform well in additive noisy condition.

This paper describes a combined behavioral techniques based on speech and signature biometrics modalities. Fusion of multiple biometric modalities for human verification performance improvement has received considerable attention. Multi-biometric systems, which consolidate information from multiple biometric sources, are gaining popularity because they are able to overcome limitations such as non-universality, noisy sensor data, large intra-user variations and susceptibility to spoof attacks that are commonly encountered in mono modal biometric systems. Soft decision level fusion based Gaussian mixture models (GMM), in which the (EM) and (GEM) algorithms for estimating the parameters of the mixture model and the number of mixture components have been compared. The test performance of the fusion, EER=0.0 % for "EM" and EER=0.02 % for "GEM", show that the combined behavioral information scheme is more robust and have a discriminating power, which can be explored for identity authentication.

Automatic speaker verification (ASV) systems are among the biometric systems used in security and telephone-based remote control applications. Recent years have witnessed an increasing trend in research on such systems. These systems usually use high dimension feature vectors and therefore involve high complexity. However, there is a general belief that many of the features used in such systems are irrelevant and redundant. So far, many methods for feature dimension reduction in these systems have been proposed, most of which are wrapper-based and thus computationally expensive since system performance is used for feature subset evaluation. This involves system training and performance evaluation for each feature subset, which is a time consuming task. In this paper, we propose a feature selection approach based on Relieff algorithm for ASV systems using support vector machine (SVM) classifiers. This method is wrapper-based but makes use of Relieff weights in order to have a lower using of system performance. Thus this method has lower complexity compared to other wrapper-based methods, can lead to 69% feature dimension reduction and has a 1.25% of Equal Error Rate (EER) for the best case that appeared in RBF kernel of SVM. The proposed method has been compared with Genetic Algorithm (GA) and Ant Colony Optimization (ACO) methods for feature selection task. Results show that the EER, number of selected features and time complexity of the proposed method is lower than these methods for different kernels of SVM.

In this paper, we present a novel object detection scheme using only local contour fragments. A sample fragment extraction method decomposes a whole contour into several parts. Then, the candidate locations of corresponding fragments in test images are detected by a modified Hausdorff distance with punishment on clutter edge regions. The most probable locations are selected by Multiclass Discriminative Field (MDF). Finally, contours of the objects can be drawn with these locations and sample fragments. Our major contributions are simplifying the MDF by an undirected graph constructed by the candidate locations and directly selecting the fragment locations by this MDF. The results on our postmark database and the ETHZ database from internet show that the proposed scheme is practicable.

Density-based spatial clustering of applications with noise (DBSCAN) is a base algorithm for density based clustering. It can find out the clusters of different shapes and sizes from a large amount of data, which is containing noise and outliers. However, it fails to handle the local density variation that exists within the cluster. Thus, a good clustering method should allow a significant density variation within the cluster because, if we go for homogeneous clustering, a large number of smaller unimportant clusters may be generated. In this paper an enhancement of DBSCAN algorithm is proposed, which detects the clusters of different shapes, sizes that differ in local density. We introduce new algorithm Dynamic Method DBSCAN (DMDBSCAN). It selects several values of the radius of a number of objects (Eps) for different densities according to a k-dist plot. For each value of Eps, DBSCAN algorithm is adopted in order to make sure that all the clusters with respect to the corresponding density are clustered. For the next process, the points that have been clustered are ignored, which avoids marking both denser areas and sparser ones as one cluster. Experimental results are obtained from artificial data sets and UCI real data sets. The final results show that our algorithm get a good results with respect to the original DBSCAN and DVBSCAN algorithms.

In the paper, we firstly recommend a new variational model for image decomposition into cartoon and texture or noise by introducing a new function in Sobolev space, in order to overcome the inconsistency between the theoretical model and numerical simulation. Secondly, we prove the existence of minimal solutions of the improved ROF energy functional. Subsequently, we also introduce two additional improved models in the same way. Finally, we show some numerical experiments of our improved ROF models, and correspondingly compare them with those of the ROF model, VO model and TV- 1 H model. The results show that our models work well.

This paper proposes a ‘Reversible Watermarking scheme for Image Authentication’ (RWIA) using Integer Wavelet Transform that satisfies the requirements of imperceptibility, capacity, and robustness. The proposed algorithm hides the data and the bookkeeping information in the high frequency subbands of CDF (2,2) integer wavelet coefficients whose magnitudes are similar to a certain predefined threshold. Histogram modification is applied as a preprocessing to prevent overflow/underflow. The embedding technique is based on the parent-child structure of the transformed coefficients called “quadruple wavelet tree” (QWT). The paper provides a comprehensive analysis of the existing methods. The experimental results evaluated by the proposed technique on different grayscale images and a comparison with existing methods is found better. The watermark is extracted to an acceptable degree of normalized cross correlation even in the presence of attacks like geometric transformations and JPEG compression.

This research needs to be held an expression introduction to the one of Yogyakarta’s masks, which is Joking Mask, using Eigen value method. Joking Mask has many expressions such as neutral, happy, sad, and angry. The detection process is by reading mask image, color, and shape, taking an important feature from the mask expression, which is on lips. Then knowing the mask expression by comparing acquired Eigen value and Eigen value from the database, so the expression can be taken as a reference to determine the most wanted expression. By using the process above, the system can recognize the Yogyakarta joking mask expression with 70% of accuracy.

With the evolution of HCI (Human-Computer Interaction), the computer vision systems are playing an important role in our lives. Some of the prime areas of computer vision applications include gender detection, face recognition, body tracking and ethnicity identification etc. Automated data analyses techniques help discover regularities and hidden associations in larger volumes of datasets. Classification being a data mining technique is largely used to group categorical data as well as a blend of continuous numeric values and categorical data. A number of classification techniques like decision trees, support vector machine (SVM), nearest neighbors and neural networks etc. have gained popularity in numerous areas of data mining practices. Among these classification techniques, decision trees offer an added advantage of producing easily interpretable rules and logic statements along with generating the classification tree for the given dataset. This study offers a distinct method for gender classification of facial images. We have used a variant of the decision tree algorithm for gender classification of frontal images due to its distinctive features. Our technique demonstrates robustness and relative scale invariance for gender classification. Details of the experimental design and the results are reported herein.

Multispectral approach to brain MRI analysis has shown great advance recently in pathology and tissue analysis. However, poor performance of the feature extraction and classification techniques involved in it discourages radiologists to use it in clinical applications. Transform based feature extraction methods like Independent Component Analysis (ICA) and its variants have contributed a lot in this research field. But these global transforms often fails in extraction of local features like small lesions from clinical cases and noisy data. Feature extraction part of the recently introduced Multiresolution Independent Component Analysis (MICA) algorithm in microarray classification is proposed in this work to resolve this issue. Effectiveness of the algorithm in MRI analysis is demonstrated by training and classification with Support Vector Machines (SVM). Both synthetic and real abnormal data from T1-weighted, T2-weighted, proton density, fluid-attenuated inversion recovery and diffusion weighted MRI sequences are considered for detailed evaluation of the method. Tanimoto index, sensitivity, specificity and accuracy of the classified results are measured and analyzed for brain abnormalities, affected white matter and gray matter tissues in all cases including noisy environment. A detailed comparative study of classification using MICA and ICA is also carried out to confirm the positive effect of the proposed method. MICA based SVM is found to yield very good results in anomaly detection, around 2.5 times improvement in classification accuracy is observed for abnormal data analysis.

The high prevalence of lung cancer, many researcher concerns about diagnosing pulmo-nary lesions in chest computed tomography (CT). However, specialists would spend a great amount of their time and effort to analysis those CT scans. And the inter-reader variability in the detection of nodules by specialists may exist. Therefore, many automated methods have proposed methods for automatic diagnosis to assist artificial inspection. This study proposes a novel hybrid method to initially classify lungs images. Firstly, adjusting the contrast of chest images can change those images from indistinct to clear, and then use the proposed novel hybrid method to automated identification CT images. From the experiments, this paper can obtain three contributions: (1) Proposed segmentation algorithm can refine the lungs regions and improve the classification performance. (2) The proposed method can be execut-ed before doctor diagnosis or computer-aided system, which can be sure that input CT image need to be detected out the actual positions, shapes or other information of nodules. (3) The results display a higher accuracy in proposed rough classifier based on DWPT-SVD than other classification methods, which verifies that proposed method can reduce time and cost of lung nodule diagnosis.

This paper presents a detailed study of human body microwave radiation by methods of stochastic dynamics. Weak electromagnetic signals are received by a Dicke radiometer and their characteristics are investigated by statistical methods to exclude possible artifacts. The signals’ deterministic content is detected by several methods of stochastic dynamics, and multi-attractors in the radiometric signals are observed. A novel method of finding the Hausdorff dimensions in the signals is proposed. The number of attractors and characteristics of their dimensions in each signal is found by means of the cluster analysis technique. These attractor dimensions of radiation of human bodies are compared with those of the background and radiometer noises, and the differences between these characteristics are noticed. Several hypotheses on the origins of this new interesting finding, including the physiological reasons, are proposed.

As a global feature of fingerprints, singular point plays important roles in fingerprint model, synthesis fingerprint, fingerprint classification, fingerprint alignment and so on. In our previous work, a rapid and effective fingerprint singular points detection method was proposed. That method detects singular points based on partitioning the orientation field into a serious non-overlapping homogenous areas. However, for the same orientation field, different partition schemes may lead to different singular point extraction results, especially for poor quality fingerprint images. For eliminating the uncertainty, we employ three different partition schemes simultaneously for each orientation field, and combine the three extraction results to make the final decision. Experiment proves that: this method is more accuracy than the previous one.

The effect of swirling on the flow characteristics of tubes with sudden expansion is experimentally examined by 3D particle image velocimetry(PIV) technique to determine velocity profiles. The swirling flow of water in sudden 1:2 axisymmetric expansion has been previously studied within a horizontal round tube. A tangential slot is used as a swirl generator for swirling flow, whereas a honey comb is used for flow without swirling. Three velocities with swirling flows are compared along the test section at Re=14800, and the streamline in the recirculation zone is described. In addition, heat transfer results are include and compared the Nusselt number with and without swirl flow for reattachment length.