Abstract: Scale Invariant Feature Transform (SIFT) has been
widely applied, but extracting SIFT feature is complicated and
time-consuming. In this paper, to meet the demand of the real-time
applications, SIFT is parallelized and optimized on cluster system,
which is named pSIFT. Redundancy storage and communication are
used for boundary data to improve the performance, and before
representation of feature descriptor, data reallocation is adopted to
keep load balance in pSIFT. Experimental results show that pSIFT
achieves good speedup and scalability.
Abstract: Ice cover County has a significant impact on rivers as it affects with the ice melting capacity which results in flooding, restrict navigation, modify the ecosystem and microclimate. River ices are made up of different ice types with varying ice thickness, so surveillance of river ice plays an important role. River ice types are captured using infrared imaging camera which captures the images even during the night times. In this paper the river ice infrared texture images are analysed using first-order statistical methods and secondorder statistical methods. The second order statistical methods considered are spatial gray level dependence method, gray level run length method and gray level difference method. The performance of the feature extraction methods are evaluated by using Probabilistic Neural Network classifier and it is found that the first-order statistical method and second-order statistical method yields low accuracy. So the features extracted from the first-order statistical method and second-order statistical method are combined and it is observed that the result of these combined features (First order statistical method + gray level run length method) provides higher accuracy when compared with the features from the first-order statistical method and second-order statistical method alone.
Abstract: Prior research evidenced that unimodal biometric
systems have several tradeoffs like noisy data, intra-class variations,
restricted degrees of freedom, non-universality, spoof attacks, and
unacceptable error rates. In order for the biometric system to be more
secure and to provide high performance accuracy, more than one
form of biometrics are required. Hence, the need arise for multimodal
biometrics using combinations of different biometric modalities. This
paper introduces a multimodal biometric system (MMBS) based on
fusion of whole dorsal hand geometry and fingerprints that acquires
right and left (Rt/Lt) near-infra-red (NIR) dorsal hand geometry (HG)
shape and (Rt/Lt) index and ring fingerprints (FP). Database of 100
volunteers were acquired using the designed prototype. The acquired
images were found to have good quality for all features and patterns
extraction to all modalities. HG features based on the hand shape
anatomical landmarks were extracted. Robust and fast algorithms for
FP minutia points feature extraction and matching were used. Feature
vectors that belong to similar biometric traits were fused using
feature fusion methodologies. Scores obtained from different
biometric trait matchers were fused using the Min-Max
transformation-based score fusion technique. Final normalized scores
were merged using the sum of scores method to obtain a single
decision about the personal identity based on multiple independent
sources. High individuality of the fused traits and user acceptability
of the designed system along with its experimental high performance
biometric measures showed that this MMBS can be considered for
med-high security levels biometric identification purposes.
Abstract: In this paper, a novel road extraction method using Stationary Wavelet Transform is proposed. To detect road features from color aerial satellite imagery, Mexican hat Wavelet filters are used by applying the Stationary Wavelet Transform in a multiresolution, multi-scale, sense and forming the products of Wavelet coefficients at a different scales to locate and identify road features at a few scales. In addition, the shifting of road features locations is considered through multiple scales for robust road extraction in the asymmetry road feature profiles. From the experimental results, the proposed method leads to a useful technique to form the basis of road feature extraction. Also, the method is general and can be applied to other features in imagery.
Abstract: In Geographic Information System, one of the sources
of obtaining needed geographic data is digitizing analog maps and
evaluation of aerial and satellite photos. In this study, a method will
be discussed which can be used to extract vectorial features and
creating vectorized drawing files for aerial photos. At the same time
a software developed for these purpose. Converting from raster to
vector is also known as vectorization and it is the most important step
when creating vectorized drawing files. In the developed algorithm,
first of all preprocessing on the aerial photo is done. These are;
converting to grayscale if necessary, reducing noise, applying some
filters and determining the edge of the objects etc. After these steps,
every pixel which constitutes the photo are followed from upper left
to right bottom by examining its neighborhood relationship and one
pixel wide lines or polylines obtained. The obtained lines have to be
erased for preventing confusion while continuing vectorization
because if not erased they can be perceived as new line, but if erased
it can cause discontinuity in vector drawing so the image converted
from 2 bit to 8 bit and the detected pixels are expressed as a different
bit. In conclusion, the aerial photo can be converted to vector form
which includes lines and polylines and can be opened in any CAD
application.
Abstract: In rotating machinery one of the critical components
that is prone to premature failure is the rolling bearing.
Consequently, early warning of an imminent bearing failure is much
critical to the safety and reliability of any high speed rotating
machines. This study is concerned with the application of Recurrence
Quantification Analysis (RQA) in fault detection of rolling element
bearings in rotating machinery. Based on the results from this study it
is reported that the RQA variable, percent determinism, is sensitive
to the type of fault investigated and therefore can provide useful
information on bearing damage in rolling element bearings.
Abstract: In this paper, an Arabic letter recognition system based on Artificial Neural Networks (ANNs) and statistical analysis for feature extraction is presented. The ANN is trained using the Least Mean Squares (LMS) algorithm. In the proposed system, each typed Arabic letter is represented by a matrix of binary numbers that are used as input to a simple feature extraction system whose output, in addition to the input matrix, are fed to an ANN. Simulation results are provided and show that the proposed system always produces a lower Mean Squared Error (MSE) and higher success rates than the current ANN solutions.
Abstract: AAM (active appearance model) has been successfully
applied to face and facial feature localization. However, its performance is sensitive to initial parameter values. In this paper, we propose a two-stage AAM for robust face alignment, which first fits an
inner face-AAM model to the inner facial feature points of the face and then localizes the whole face and facial features by optimizing the
whole face-AAM model parameters. Experiments show that the proposed face alignment method using two-stage AAM is more reliable to the background and the head pose than the standard
AAM-based face alignment method.
Abstract: Many factors affect the success of Machine Learning
(ML) on a given task. The representation and quality of the instance
data is first and foremost. If there is much irrelevant and redundant
information present or noisy and unreliable data, then knowledge
discovery during the training phase is more difficult. It is well known
that data preparation and filtering steps take considerable amount of
processing time in ML problems. Data pre-processing includes data
cleaning, normalization, transformation, feature extraction and
selection, etc. The product of data pre-processing is the final training
set. It would be nice if a single sequence of data pre-processing
algorithms had the best performance for each data set but this is not
happened. Thus, we present the most well know algorithms for each
step of data pre-processing so that one achieves the best performance
for their data set.
Abstract: Brain Computer Interface (BCI) has been recently
increased in research. Functional Near Infrared Spectroscope (fNIRs)
is one the latest technologies which utilize light in the near-infrared
range to determine brain activities. Because near infrared technology
allows design of safe, portable, wearable, non-invasive and wireless
qualities monitoring systems, fNIRs monitoring of brain
hemodynamics can be value in helping to understand brain tasks. In
this paper, we present results of fNIRs signal analysis indicating that
there exist distinct patterns of hemodynamic responses which
recognize brain tasks toward developing a BCI. We applied two
different mathematics tools separately, Wavelets analysis for
preprocessing as signal filters and feature extractions and Neural
networks for cognition brain tasks as a classification module. We
also discuss and compare with other methods while our proposals
perform better with an average accuracy of 99.9% for classification.
Abstract: This paper presents the automated methods employed
for extracting craniofacial landmarks in white light images as part of
a registration framework designed to support three neurosurgical
procedures. The intraoperative space is characterised by white light
stereo imaging while the preoperative plan is performed on CT scans.
The registration aims at aligning these two modalities to provide a
calibrated environment to enable image-guided solutions. The
neurosurgical procedures can then be carried out by mapping the
entry and target points from CT space onto the patient-s space. The
registration basis adopted consists of natural landmarks (eye corner
and ear tragus). A 5mm accuracy is deemed sufficient for these three
procedures and the validity of the selected registration basis in
achieving this accuracy has been assessed by simulation studies. The
registration protocol is briefly described, followed by a presentation
of the automated techniques developed for the extraction of the
craniofacial features and results obtained from tests on the AR and
FERET databases. Since the three targeted neurosurgical procedures
are routinely used for head injury management, the effect of
bruised/swollen faces on the automated algorithms is assessed. A
user-interactive method is proposed to deal with such unpredictable
circumstances.
Abstract: This paper proposes a neural network weights and
topology optimization using genetic evolution and the
backpropagation training algorithm. The proposed crossover and
mutation operators aims to adapt the networks architectures and
weights during the evolution process. Through a specific inheritance
procedure, the weights are transmitted from the parents to their
offsprings, which allows re-exploitation of the already trained
networks and hence the acceleration of the global convergence of the
algorithm. In the preprocessing phase, a new feature extraction
method is proposed based on Legendre moments with the Maximum
entropy principle MEP as a selection criterion. This allows a global
search space reduction in the design of the networks. The proposed
method has been applied and tested on the well known MNIST
database of handwritten digits.
Abstract: In this paper, a new face recognition method based on
PCA (principal Component Analysis), LDA (Linear Discriminant
Analysis) and neural networks is proposed. This method consists of
four steps: i) Preprocessing, ii) Dimension reduction using PCA, iii)
feature extraction using LDA and iv) classification using neural
network. Combination of PCA and LDA is used for improving the
capability of LDA when a few samples of images are available and
neural classifier is used to reduce number misclassification caused by
not-linearly separable classes. The proposed method was tested on
Yale face database. Experimental results on this database
demonstrated the effectiveness of the proposed method for face
recognition with less misclassification in comparison with previous
methods.
Abstract: In this paper we propose a method for recognition of
adult video based on support vector machine (SVM). Different kernel
features are proposed to classify adult videos. SVM has an advantage
that it is insensitive to the relative number of training example in
positive (adult video) and negative (non adult video) classes. This
advantage is illustrated by comparing performance between different
SVM kernels for the identification of adult video.
Abstract: Classification is an interesting problem in functional
data analysis (FDA), because many science and application problems
end up with classification problems, such as recognition, prediction,
control, decision making, management, etc. As the high dimension
and high correlation in functional data (FD), it is a key problem to
extract features from FD whereas keeping its global characters, which
relates to the classification efficiency and precision to heavens. In this
paper, a novel automatic method which combined Genetic Algorithm
(GA) and classification algorithm to extract classification features is
proposed. In this method, the optimal features and classification model
are approached via evolutional study step by step. It is proved by
theory analysis and experiment test that this method has advantages in
improving classification efficiency, precision and robustness whereas
using less features and the dimension of extracted classification
features can be controlled.
Abstract: This paper presents a comparison of metaheuristic
algorithms, Genetic Algorithm (GA) and Ant Colony Optimization
(ACO), in producing freeman chain code (FCC). The main problem
in representing characters using FCC is the length of the FCC
depends on the starting points. Isolated characters, especially the
upper-case characters, usually have branches that make the traversing
process difficult. The study in FCC construction using one
continuous route has not been widely explored. This is our
motivation to use the population-based metaheuristics. The
experimental result shows that the route length using GA is better
than ACO, however, ACO is better in computation time than GA.
Abstract: This paper reports a new pattern recognition approach for face recognition. The biological model of light receptors - cones and rods in human eyes and the way they are associated with pattern vision in human vision forms the basis of this approach. The functional model is simulated using CWD and WPD. The paper also discusses the experiments performed for face recognition using the features extracted from images in the AT & T face database. Artificial Neural Network and k- Nearest Neighbour classifier algorithms are employed for the recognition purpose. A feature vector is formed for each of the face images in the database and recognition accuracies are computed and compared using the classifiers. Simulation results show that the proposed method outperforms traditional way of feature extraction methods prevailing for pattern recognition in terms of recognition accuracy for face images with pose and illumination variations.
Abstract: In this paper, we propose effective system for digital music retrieval. We divided proposed system into Client and Server. Client part consists of pre-processing and Content-based feature extraction stages. In pre-processing stage, we minimized Time code Gap that is occurred among same music contents. As content-based feature, first-order differentiated MFCC were used. These presented approximately envelop of music feature sequences. Server part included Music Server and Music Matching stage. Extracted features from 1,000 digital music files were stored in Music Server. In Music Matching stage, we found retrieval result through similarity measure by DTW. In experiment, we used 450 queries. These were made by mixing different compression standards and sound qualities from 50 digital music files. Retrieval accurate indicated 97% and retrieval time was average 15ms in every single query. Out experiment proved that proposed system is effective in retrieve digital music and robust at various user environments of web.
Abstract: Image retrieval is a topic where scientific interest is currently high. The important steps associated with image retrieval system are the extraction of discriminative features and a feasible similarity metric for retrieving the database images that are similar in content with the search image. Gabor filtering is a widely adopted technique for feature extraction from the texture images. The recently proposed sparsity promoting l1-norm minimization technique finds the sparsest solution of an under-determined system of linear equations. In the present paper, the l1-norm minimization technique as a similarity metric is used in image retrieval. It is demonstrated through simulation results that the l1-norm minimization technique provides a promising alternative to existing similarity metrics. In particular, the cases where the l1-norm minimization technique works better than the Euclidean distance metric are singled out.
Abstract: The applications on numbers are across-the-board that there is much scope for study. The chic of writing numbers is diverse and comes in a variety of form, size and fonts. Identification of Indian languages scripts is challenging problems. In Optical Character Recognition [OCR], machine printed or handwritten characters/numerals are recognized. There are plentiful approaches that deal with problem of detection of numerals/character depending on the sort of feature extracted and different way of extracting them. This paper proposes a recognition scheme for handwritten Hindi (devnagiri) numerals; most admired one in Indian subcontinent our work focused on a technique in feature extraction i.e. Local-based approach, a method using 16-segment display concept, which is extracted from halftoned images & Binary images of isolated numerals. These feature vectors are fed to neural classifier model that has been trained to recognize a Hindi numeral. The archetype of system has been tested on varieties of image of numerals. Experimentation result shows that recognition rate of halftoned images is 98 % compared to binary images (95%).