Fault detection in electrical equipment’s images by using optimal features with deep learning classifier

Multimedia Tools and Applications - Infrared imaging frameworks have been broadly utilized as a part of the military and civil fields, for example, target recognition, fault diagnosis, fire...     

Diagnosis of lung nodule using Moran’s index and Geary’s coefficient in computerized tomography images

This paper analyzes the application of Moran’s index and Geary’s coefficient to the characterization of lung nodules as malignant or benign in computerized tomography images. The characterization method is based on a process that verifies which combination of measures, from the proposed measures, has been best able to discriminate between the benign and malignant nodules using stepwise discriminant analysis. Then, a linear discriminant analysis procedure was performed using the selected features to evaluate the ability of these in predicting the classification for each nodule. In order to verify this application we also describe tests that were carried out using a sample of 36 nodules: 29 benign and 7 malignant. A leave-one-out procedure was used to provide a less biased estimate of the linear discriminator’s performance. The two analyzed functions and its combinations have provided above 90% of accuracy and a value area under receiver operation characteristic (ROC) curve above 0.85, that indicates a promising potential to be used as nodules signature measures. The preliminary results of this approach are very encouraging in characterizing nodules using the two functions presented.

Driver’s eye blinking detection using novel color and texture segmentation algorithms

In this paper we propose a system that measures eye blinking rate and eye closure duration. The system consists of skin-color segmentation, facial features segmentation, iris positioning and blink detection. The proposed skin-segmentation procedure is based on a neural network approximation of a RGB skin-color histogram. This method is robust and adaptive to any skin-color training set. The largest remaining skin-color region among skin-color segmentation results is further segmented into open/closed eyes, lips, nose, eyebrows, and the remaining facial regions using a novel texture segmentation algorithm. The segmentation algorithm classifies pixels according to the highest probability among the estimated facial feature class probability density functions (PDFs). The segmented eye regions are analyzed with the Circular Hough transform with the purpose of finding iris candidates. The finial iris position is selected according to the location of the maximum correlation value obtained from correlation with a predefined mask. The positions of irises and eye states are monitored through time to estimate eye blinking frequency and eye closure duration. The method of the driver drowsiness detection using these parameters is illustrated. The proposed system is tested on CCD and CMOS cameras under different environmental conditions and the experimental results show high system performance.     

Integration of panchromatic and SAR features into multispectral SPOT images using the ‘à trous’ wavelet decomposition

A method is described for integrating panchromatic (P) and synthetic aperture radar (SAR) features into multispectral (XS) images using conjointly the modified Brovey transform (MBT) and the ‘à trous’ wavelet decomposition (ATDW). The MBT is based on the local modulation of each multispectral image by the ratio of the new and initial intensity components to produce new multispectral images directly. The ATWD allows extraction of features from P and SAR images, which are combined through a feature selection rule to integrate into the initial intensity component. For evaluating the effect of each feature selection on new XS images, experimental results are conducted on SPOT (XS, P) and Radarsat (SAR) images using both visual inspection and many refined statistical measures.     

Spam detection on Twitter using a support vector machine and users’ features by identifying their interactions

Multimedia Tools and Applications - Spam tweets might cause numerous problems for users. An automatic method is introduced as a proposed method to detect spam tweets. This method is based on...     

A novel Parkinson’s Disease Diagnosis Index using higher-order spectra features in EEG signals

Higher-order spectra (HOS) is an efficient feature extraction method used in various biomedical applications such as stages of sleep, epilepsy detection, cardiac abnormalities, and affective computing. The motive of this work was to explore the application of HOS for an automated diagnosis of Parkinson’s disease (PD) using electroencephalography (EEG) signals. Resting-state EEG signals collected from 20 PD patients with medication and 20 age-matched normal subjects were used in this study. HOS bispectrum features were extracted from the EEG signals. The obtained features were ranked using t value, and highly ranked features were used in order to develop the PD Diagnosis Index (PDDI). The PDDI is a single value, which can discriminate the two classes. Also, the ranked features were fed one by one to the various classifiers, namely decision tree (DT), fuzzy K-nearest neighbor (FKNN), K-nearest neighbor (KNN), naive bayes (NB), probabilistic neural network (PNN), and support vector machine (SVM), to choose the best classifier using minimum number of features. We have obtained an optimum mean classification accuracy of 99.62%, mean sensitivity and specificity of 100.00 and 99.25%, respectively, using the SVM classifier. The proposed PDDI can aid the clinicians in their diagnosis and help to test the efficacy of drugs.

     

Early detection of Alzheimer’s disease using local binary pattern and convolutional neural network

Multimedia Tools and Applications - Alzheimer’s disease, a progressive and irreversible abnormality of the human brain impairs memory and thinking skills. Gradually, it will damage the...     

Texture features’ based classification of MR images of normal and herniated intervertebral discs

Multimedia Tools and Applications - Disc herniation is considered as a very common spine abnormality resulting in severe pain in back and legs. Besides it has great impact on economy of suffering...     

Shadow detection in colour high‐resolution satellite images

Image shadow segmentation has become a major issue in satellite remote sensing because of the recent commercial availability of high‐resolution images. Detecting shadows is important for successfully carrying out applications such as change detection, land monitoring, object recognition, scene reconstruction, colour correction, etc. This paper presents a simple and effective procedure to segment shadow regions on high‐resolution colour satellite images. The method applies a region growing process on a specific band (namely, the c ₃ component of the c ₁ c ₂ c ₃ colour space). To gain in robustness and precision, the region expansion also imposes a restriction on the saturation and intensity values of the shadow pixels, as well as on their edge gradients. The proposed method has been successfully tested on QuickBird images acquired under different lighting conditions and covering both urban and rural areas.     

Evaluating Algeria’s social and economic development using a series of night-time light images between 1992 to 2012

ABSTRACT

After social and economic problems, Algeria regained its night-time light (NL) brightness as this analysis of space-borne images shows. We evaluate the advancement of artificial light sources and the socioeconomic development between 1992 to 2012 for all 48 provinces. Image calibration improved the consistency and comparability of the DMSP-OLS data. Light detected from gas flaring has been separated to focus on light from human settlements. We then estimated the relationship between night-time light and selected socioeconomic parameters (SEP). Night-time light growth (NLG) and night-time light annual growth (NLAG) indices are calculated at the national, regional and provincial level. Considering the Oil price crash and the Algerian Civil Concord, we take a detailed look at these historic events in Algeria. Our main findings are: (A) The total night-time light (TNL) for human settlements grew nationwide by 128%, (B) night-time light data correlates with socioeconomic parameters such as population, and electric power consumption, (C) the oil price crash resulted in slower night-time light growth, while the Algerian Civil Concord and the end of the crises led to increased TNL, (D) because of government investment, the High Plateaus are the fastest growing region since the end of the crises, (E) space-borne night-time light data can consistently fill the information gap when data from official sources are not available. Government and policy-making bodies may benefit from the presented findings allowing them to oversee and address provincial inequalities and economic development.     

Zebra mussels’ behaviour detection,extraction and classification using wavelets and kernel methods

This paper concerns the detection, feature extraction and classification of behaviours of Dreissena polymorpha. A new algorithm based on wavelets and kernel methods that detects relevant events in the collected data is presented. This algorithm allows us to extract elementary events from the behaviour of a living organism. Moreover, we propose an efficient framework for automatic classification to separate the control and stressful conditions.     

Prediction of lysine propionylation sites using biased SVM and incorporating four different sequence features into Chou’s PseAAC

Lysine propionylation is an important and common protein acylation modification in both prokaryotes and eukaryotes. To better understand the molecular mechanism of propionylation, it is important to identify propionylated substrates and their corresponding propionylation sites accurately. In this study, a novel bioinformatics tool named PropPred is developed to predict propionylation sites by using multiple feature extraction and biased support vector machine. On the one hand, various features are incorporated, including amino acid composition, amino acid factors, binary encoding, and the composition of k-spaced amino acid pairs. And the F-score feature method and the incremental feature selection algorithm are adopted to remove the redundant features. On the other hand, the biased support vector machine algorithm is used to handle the imbalanced problem in propionylation sites training dataset. As illustrated by 10-fold cross-validation, the performance of PropPred achieves a satisfactory performance with a Sensitivity of 70.03%, a Specificity of 75.61%, an accuracy of 75.02% and a Matthew’s correlation coefficient of 0.3085. Feature analysis shows that some amino acid factors play the most important roles in the prediction of propionylation sites. These analysis and prediction results might provide some clues for understanding the molecular mechanisms of propionylation. A user-friendly web-server for PropPred is established at 123.206.31.171/PropPred/.     

Optimizing player’s formations for corner-kick situations in RoboCup soccer 2D simulation

In the domain of the Soccer simulation 2D league of the RoboCup, appropriate player positioning against the opponent team formation is an important factor of soccer team performance. In this work, we propose to use a meta-heuristic algorithm called the firefly algorithm to optimize player positioning. We used sequential Bayesian estimation as well as parallelization to reduce the necessary number of time-consuming simulated soccer matches. As a first trial of our system, we optimized the corner-kick formation. Preliminary results in optimizing the corner-kick formation are not advantageous over the previous handmade formation due to the difficulty in tuning the meta-heuristic algorithm parameters. However, it is also shown that the proposed system is effective in handling a high load of simulations over the span of weeks and therefore is promising to be usable to optimize player positioning.     

Image indexing and content analysis in children’s picture books using a large-scale database

In this paper we introduce a visual database for children’s picture book and we also present an intelligent robot trained on this database. Firstly, a large-scale image dataset is built that contains image samples of book pages. It can be used to verify image indexing algorithms and content recognition algorithms. Secondly, we study the state-of-the-art algorithms in image matching and object recognition. Several approaches are presented and compared from the aspects of computational efficiency and recognition accuracy. In order to improve the speed we proposed a novel hierarchical algorithm for fast search. Finally, using this large-scale database we are able to build a robot that can read children’s picture books and initial experimental results are presented. We can see that both the training database and the algorithms are promising, yet there are still a few open challenges concerning the costs and robustness.

     

Fewer is more:efficient object detection in large aerial images

Current mainstream object detection methods for large aerial images usually divide large images into patches and then exhaustively detect the objects of interest on all patches,no matter whether there exist objects or not.This paradigm,although effective,is inefficient because the detectors have to go through all patches,severely hindering the inference speed.This paper presents an objectness activation network(OAN) to help detectors focus on fewer patches but achieve more efficient inference an...     

Predicting user’s preferences using neural networks and psychology models

In this paper we describe a new model suitable for optimization problems with explicitly unknown optimization functions using user’s preferences. The model addresses an ability to learn not known optimization functions thus perform also a learning of user’s preferences. The model consists of neural networks using fuzzy membership functions and interactive evolutionary algorithms in the process of learning. Fuzzy membership functions of basic human values and their priorities were prepared by utilizing Schwartz’s model of basic human values (achievement, benevolence, conformity, hedonism, power, security, self-direction, stimulation, tradition and universalism). The quality of the model was tested on “the most attractive font face problem” and it was evaluated using the following criteria: a speed of optimal parameters computation, a precision of achieved results, Wilcoxon signed rank test and a similarity of letter images. The results qualify the developed model as very usable in user’s preference modeling.     

Fluid flow-based description of the geometrical features in fluidic channels using the Shannon’s information theory: an exploratory study

Microfluidic analytical devices manufactured on paper and similar inexpensive substrates (µ-PADs) have shown considerable promise for disease diagnostics in resource-limited regions. However, current commercialization approaches can be improved substantially by addressing existing technical challenges associated with µ-PADs. Among these, off-device plasma separation from whole blood is a critical challenge in µ-PAD technology that limits commercialization. Existing µ-PADs made by combining multiple components require extra fabrication steps and manufacturing material. Our approach utilizes a two-step plasma process to fabricate single-layer semi-enclosed µ-PADs directly on a commercially available blood plasma separation membrane to incorporate blood plasma separation functionality into the device. The semi-enclosed µ-PADs are bonded with low-cost adhesive plastic tape to provide mechanical support to the device and make it more mechanically robust for field applications. Detection zones of the µ-PADs have also been modified with a cellulose nanocrystal (CNC) to increase colorimetric signal homogeneity, thus enhancing signal quality. The CNC-modified µ-PADs have been used for colorimetric detection of two model analytes (glucose and albumin) in whole blood. Colorimetric signals for both glucose and albumin from whole blood samples were consistent with the calibration curves generated using stock solutions.

     

Hyperspectral anomaly detection using spectral–spatial features based on the human visual system

ABSTRACT

Anomaly detection (AD) is one of the most attracting topics within the recent 10 years in hyperspectral imagery (HSI). The goal of the AD is to label the pixels with significant spectral or spatial differences to their neighbours, as targets. In this paper, we propose a method that uses both spectral and spatial information of HSI based on human visual system (HVS). By inspiring the retina and the visual cortex functionality, the multiscale multiresolution analysis is applied to some principal components of hyperspectral data, to extract features from different spatial levels of the image. Then the global and local relations between features are considered based on inspiring the visual attention mechanism and inferotemporal (IT) part of the visual cortex. The effects of the attention mechanism are implemented using the logarithmic function which well highlights, small variations in pixels’ grey levels in global features. Also, the maximum operation is used over the local features for imitating the function of IT. Finally, the information theory concept is used for generating the final detection map by weighting the global and local detection maps to obtain the final anomaly map. The result of the proposed method is compared with some state-of-the-art methods such as SSRAD, FLD, PCA, RX, KPCA, and AED for two well-known real hyperspectral data which are San Diego airport and Pavia city, and a synthetic hyperspectral data. The results demonstrate that the proposed method effectively improves the AD capabilities, such as enhancement of the detection rate, reducing the false alarm rate and the computation complexity.