Abstract: This paper describes new computer vision algorithms
that have been developed to track moving objects as part of a
long-term study into the design of (semi-)autonomous vehicles. We
present the results of a study to exploit variable kernels for tracking in
video sequences. The basis of our work is the mean shift
object-tracking algorithm; for a moving target, it is usual to define a
rectangular target window in an initial frame, and then process the data
within that window to separate the tracked object from the background
by the mean shift segmentation algorithm. Rather than use the
standard, Epanechnikov kernel, we have used a kernel weighted by the
Chamfer distance transform to improve the accuracy of target
representation and localization, minimising the distance between the
two distributions in RGB color space using the Bhattacharyya
coefficient. Experimental results show the improved tracking
capability and versatility of the algorithm in comparison with results
using the standard kernel. These algorithms are incorporated as part of
a robot test-bed architecture which has been used to demonstrate their
effectiveness.
Abstract: Support vector machines (SVMs) have shown
superior performance compared to other machine learning techniques,
especially in classification problems. Yet one limitation of SVMs is
the lack of an explanation capability which is crucial in some
applications, e.g. in the medical and security domains. In this paper, a
novel approach for eclectic rule-extraction from support vector
machines is presented. This approach utilizes the knowledge acquired
by the SVM and represented in its support vectors as well as the
parameters associated with them. The approach includes three stages;
training, propositional rule-extraction and rule quality evaluation.
Results from four different experiments have demonstrated the value
of the approach for extracting comprehensible rules of high accuracy
and fidelity.
Abstract: A generalized Digital Modulation Identification algorithm for adaptive demodulator has been developed and presented in this paper. The algorithm developed is verified using wavelet Transform and histogram computation to identify QPSK and QAM with GMSK and M–ary FSK modulations. It has been found that the histogram peaks simplifies the procedure for identification. The simulated results show that the correct modulation identification is possible to a lower bound of 5 dB and 12 dB for GMSK and QPSK respectively. When SNR is above 5 dB the throughput of the proposed algorithm is more than 97.8%. The receiver operating characteristics (ROC) has been computed to measure the performance of the proposed algorithm and the analysis shows that the probability of detection (Pd) drops rapidly when SNR is 5 dB and probability of false alarm (Pf) is smaller than 0.3. The performance of the proposed algorithm has been compared with existing methods and found it will identify all digital modulation schemes with low SNR.
Abstract: Video Mosaicing is the stitching of selected frames of
a video by estimating the camera motion between the frames and
thereby registering successive frames of the video to arrive at the
mosaic. Different techniques have been proposed in the literature for
video mosaicing. Despite of the large number of papers dealing with
techniques to generate mosaic, only a few authors have investigated
conditions under which these techniques generate good estimate of
motion parameters. In this paper, these techniques are studied under
different videos, and the reasons for failures are found. We propose
algorithms with incorporation of outlier removal algorithms for better
estimation of motion parameters.
Abstract: Automatic detection of bleeding is of practical
importance since capsule endoscopy produces an extremely large
number of images. Algorithm development of bleeding detection in
the digestive tract is difficult due to different contrasts among the
images, food dregs, secretion and others. In this study, were assigned
weighting factors derived from the independent features of the
contrast and brightness between bleeding and normality. Spectral
analysis based on weighting factors was fast and accurate. Results
were a sensitivity of 87% and a specificity of 90% when the accuracy
was determined for each pixel out of 42 endoscope images.
Abstract: Feature and model selection are in the center of
attention of many researches because of their impact on classifiers-
performance. Both selections are usually performed separately but
recent developments suggest using a combined GA-SVM approach to
perform them simultaneously. This approach improves the
performance of the classifier identifying the best subset of variables
and the optimal parameters- values. Although GA-SVM is an
effective method it is computationally expensive, thus a rough
method can be considered. The paper investigates a joined approach
of Genetic Algorithm and kernel matrix criteria to perform
simultaneously feature and model selection for SVM classification
problem. The purpose of this research is to improve the classification
performance of SVM through an efficient approach, the Kernel
Matrix Genetic Algorithm method (KMGA).
Abstract: Motion estimation is a key problem in video
processing and computer vision. Optical flow motion estimation can
achieve high estimation accuracy when motion vector is small.
Three-step search algorithm can handle large motion vector but not
very accurate. A joint algorithm was proposed in this paper to
achieve high estimation accuracy disregarding whether the motion
vector is small or large, and keep the computation cost much lower
than full search.
Abstract: Text categorization techniques are widely used to many Information Retrieval (IR) applications. In this paper, we proposed a simple but efficient method that can automatically find the relationship between any pair of terms and documents, also an indexing matrix is established for text categorization. We call this method Indexing Matrix Categorization Machine (IMCM). Several experiments are conducted to show the efficiency and robust of our algorithm.
Abstract: In this paper, we propose a novel concept of relative
distance measurement using Stereo Vision Technology and discuss
its implementation on a FPGA based real-time image processor. We
capture two images using two CCD cameras and compare them.
Disparity is calculated for each pixel using a real time dense disparity
calculation algorithm. This algorithm is based on the concept of
indexed histogram for matching. Disparity being inversely
proportional to distance (Proved Later), we can thus get the relative
distances of objects in front of the camera. The output is displayed on
a TV screen in the form of a depth image (optionally using pseudo
colors). This system works in real time on a full PAL frame rate (720
x 576 active pixels @ 25 fps).
Abstract: This paper describes a UDP over IP based, server-oriented redundant host configuration protocol (RHCP) that can be used by collaborating embedded systems in an ad-hoc network to acquire a dynamic IP address. The service is provided by a single network device at a time and will be dynamically reassigned to one of the other network clients if the primary provider fails. The protocol also allows all participating clients to monitor the dynamic makeup of the network over time. So far the algorithm has been implemented and tested on an 8-bit embedded system architecture with a 10Mbit Ethernet interface.
Abstract: In this paper we present a technique to speed up
ICA based on the idea of reducing the dimensionality of the data
set preserving the quality of the results. In particular we refer to
FastICA algorithm which uses the Kurtosis as statistical property
to be maximized. By performing a particular Johnson-Lindenstrauss
like projection of the data set, we find the minimum dimensionality
reduction rate ¤ü, defined as the ratio between the size k of the reduced
space and the original one d, which guarantees a narrow confidence
interval of such estimator with high confidence level. The derived
dimensionality reduction rate depends on a system control parameter
β easily computed a priori on the basis of the observations only.
Extensive simulations have been done on different sets of real world
signals. They show that actually the dimensionality reduction is very
high, it preserves the quality of the decomposition and impressively
speeds up FastICA. On the other hand, a set of signals, on which the
estimated reduction rate is greater than 1, exhibits bad decomposition
results if reduced, thus validating the reliability of the parameter β.
We are confident that our method will lead to a better approach to
real time applications.
Abstract: Electro-optical devices are increasingly used for
military sea-, land- and air applications to detect, recognize and track
objects. Typically, these devices produce video information that is
presented to an operator. However, with increasing availability of
electro-optical devices the data volume is becoming very large,
creating a rising need for automated analysis. In a military setting,
this typically involves detecting and recognizing objects at a large
distance, i.e. when they are difficult to distinguish from background
and noise. One may consider combining multiple images from a
video stream into a single enhanced image that provides more
information for the operator. In this paper we investigate a simple
algorithm to enhance simulated images from a military context and
investigate how the enhancement is affected by various types of
disturbance.
Abstract: One of the disadvantages of using OFDM is the larger
peak to averaged power ratio (PAPR) in its time domain signal. The
larger PAPR signal would course the fatal degradation of bit error
rate performance (BER) due to the inter-modulation noise in the nonlinear
channel. This paper proposes an improved DSI (Dummy
Sequence Insertion) method, which can achieve the better PAPR and
BER performances. The feature of proposed method is to optimize
the phase of each dummy sub-carrier so as to reduce the PAPR
performance by changing all predetermined phase coefficients in the
time domain signal, which is calculated for data sub-carriers and
dummy sub-carriers separately. To achieve the better PAPR
performance, this paper also proposes to employ the time-frequency
domain swapping algorithm for fine adjustment of phase coefficient
of the dummy subcarriers, which can achieve the less complexity of
processing and achieves the better PAPR and BER performances
than those for the conventional DSI method. This paper presents
various computer simulation results to verify the effectiveness of
proposed method as comparing with the conventional methods in the
non-linear channel.
Abstract: This article presents the results using a parametric approach and a Wavelet Transform in analysing signals emitting from the sperm whale. The extraction of intrinsic characteristics of these unique signals emitted by marine mammals is still at present a difficult exercise for various reasons: firstly, it concerns non-stationary signals, and secondly, these signals are obstructed by interfering background noise. In this article, we compare the advantages and disadvantages of both methods: Auto Regressive models and Wavelet Transform. These approaches serve as an alternative to the commonly used estimators which are based on the Fourier Transform for which the hypotheses necessary for its application are in certain cases, not sufficiently proven. These modern approaches provide effective results particularly for the periodic tracking of the signal's characteristics and notably when the signal-to-noise ratio negatively effects signal tracking. Our objectives are twofold. Our first goal is to identify the animal through its acoustic signature. This includes recognition of the marine mammal species and ultimately of the individual animal (within the species). The second is much more ambitious and directly involves the intervention of cetologists to study the sounds emitted by marine mammals in an effort to characterize their behaviour. We are working on an approach based on the recordings of marine mammal signals and the findings from this data result from the Wavelet Transform. This article will explore the reasons for using this approach. In addition, thanks to the use of new processors, these algorithms once heavy in calculation time can be integrated in a real-time system.
Abstract: Unlike this study focused extensively on trading
behavior of option market, those researches were just taken their
attention to model-driven option pricing. For example, Black-Scholes
(B-S) model is one of the most famous option pricing models.
However, the arguments of B-S model are previously mentioned by
some pricing models reviewing. This paper following suggests the
importance of the dynamic character for option pricing, which is also
the reason why using the genetic algorithm (GA). Because of its
natural selection and species evolution, this study proposed a hybrid
model, the Genetic-BS model which combining GA and B-S to
estimate the price more accurate. As for the final experiments, the
result shows that the output estimated price with lower MAE value
than the calculated price by either B-S model or its enhanced one,
Gram-Charlier garch (G-C garch) model. Finally, this work would
conclude that the Genetic-BS pricing model is exactly practical.
Abstract: Video sensor networks operate on stringent requirements
of latency. Packets have a deadline within which they have
to be delivered. Violation of the deadline causes a packet to be
treated as lost and the loss of packets ultimately affects the quality
of the application. Network latency is typically a function of many
interacting components. In this paper, we propose ways of reducing
the forwarding latency of a packet at intermediate nodes. The
forwarding latency is caused by a combination of processing delay
and queueing delay. The former is incurred in order to determine the
next hop in dynamic routing. We show that unless link failures in a
very specific and unlikely pattern, a vast majority of these lookups
are redundant. To counter this we propose source routing as the
routing strategy. However, source routing suffers from issues related
to scalability and being impervious to network dynamics. We propose
solutions to counter these and show that source routing is definitely
a viable option in practical sized video networks. We also propose a
fast and fair packet scheduling algorithm that reduces queueing delay
at the nodes. We support our claims through extensive simulation on
realistic topologies with practical traffic loads and failure patterns.
Abstract: The goal of this work is to describe a new algorithm for finding the optimal variable order, number of nodes for any order and other ROBDD parameters, based on a tabular method. The tabular method makes use of a pre-built backend database table that stores the ROBDD size for selected combinations of min-terms. The user uses the backend table and the proposed algorithm to find the necessary ROBDD parameters, such as best variable order, number of nodes etc. Experimental results on benchmarks are given for this technique.
Abstract: The controllable electrical loss which consists of the
copper loss and iron loss can be minimized by the optimal control of
the armature current vector. The control algorithm of current vector
minimizing the electrical loss is proposed and the optimal current
vector can be decided according to the operating speed and the load
conditions. The proposed control algorithm is applied to the
experimental PM motor drive system and this paper presents a
modern approach of speed control for permanent magnet
synchronous motor (PMSM) applied for Electric Vehicle using a
nonlinear control. The regulation algorithms are based on the
feedback linearization technique. The direct component of the current
is controlled to be zero which insures the maximum torque operation.
The near unity power factor operation is also achieved. More over,
among EV-s motor electric propulsion features, the energy efficiency
is a basic characteristic that is influenced by vehicle dynamics and
system architecture. For this reason, the EV dynamics are taken into
account.
Abstract: In this paper, a new robust audio fingerprinting
algorithm in MP3 compressed domain is proposed with high
robustness to time scale modification (TSM). Instead of simply
employing short-term information of the MP3 stream, the new
algorithm extracts the long-term features in MP3 compressed domain
by using the modulation frequency analysis. Our experiment has
demonstrated that the proposed method can achieve a hit rate of
above 95% in audio retrieval and resist the attack of 20% TSM. It has
lower bit error rate (BER) performance compared to the other
algorithms. The proposed algorithm can also be used in other
compressed domains, such as AAC.
Abstract: Gesture recognition is a challenging task for extracting
meaningful gesture from continuous hand motion. In this paper, we propose an automatic system that recognizes isolated gesture,
in addition meaningful gesture from continuous hand motion for Arabic numbers from 0 to 9 in real-time based on Hidden Markov Models (HMM). In order to handle isolated gesture, HMM using
Ergodic, Left-Right (LR) and Left-Right Banded (LRB) topologies is applied over the discrete vector feature that is extracted from stereo
color image sequences. These topologies are considered to different
number of states ranging from 3 to 10. A new system is developed to recognize the meaningful gesture based on zero-codeword detection
with static velocity motion for continuous gesture. Therefore, the
LRB topology in conjunction with Baum-Welch (BW) algorithm for
training and forward algorithm with Viterbi path for testing presents the best performance. Experimental results show that the proposed system can successfully recognize isolated and meaningful gesture and achieve average rate recognition 98.6% and 94.29% respectively.