Abstract: This paper proposes a method for speckle reduction in
medical ultrasound imaging while preserving the edges with the
added advantages of adaptive noise filtering and speed. A nonlinear
image diffusion method that incorporates local image parameter,
namely, scatterer density in addition to gradient, to weight the
nonlinear diffusion process, is proposed. The method was tested for
the isotropic case with a contrast detail phantom and varieties of
clinical ultrasound images, and then compared to linear and some
other diffusion enhancement methods. Different diffusion parameters
were tested and tuned to best reduce speckle noise and preserve
edges. The method showed superior performance measured both
quantitatively and qualitatively when incorporating scatterer density
into the diffusivity function. The proposed filter can be used as a
preprocessing step for ultrasound image enhancement before
applying automatic segmentation, automatic volumetric calculations,
or 3D ultrasound volume rendering.
Abstract: Image target detection and tracking methods based on
target information such as intensity, shape model, histogram and
target dynamics have been proven to be robust to target model
variations and background clutters as shown by recent researches.
However, no definitive answer has been given to occluded target by
counter measure or limited field of view(FOV). In this paper, we
will present a novel tracking method using filtering and computational
geometry. This paper has two central goals: 1) to deal with vulnerable
target measurements; and 2) to maintain target tracking out of FOV
using non-target-originated information. The experimental results,
obtained with airborne images, show a robust tracking ability with
respect to the existing approaches. In exploring the questions of target
tracking, this paper will be limited to consideration of airborne image.
Abstract: Robots- visual perception is a field that is gaining
increasing attention from researchers. This is partly due to emerging
trends in the commercial availability of 3D scanning systems or
devices that produce a high information accuracy level for a variety of
applications. In the history of mining, the mortality rate of mine workers
has been alarming and robots exhibit a great deal of potentials to
tackle safety issues in mines. However, an effective vision system
is crucial to safe autonomous navigation in underground terrains.
This work investigates robots- perception in underground terrains
(mines and tunnels) using statistical region merging (SRM) model.
SRM reconstructs the main structural components of an imagery
by a simple but effective statistical analysis. An investigation is
conducted on different regions of the mine, such as the shaft, stope
and gallery, using publicly available mine frames, with a stream of
locally captured mine images. An investigation is also conducted on a
stream of underground tunnel image frames, using the XBOX Kinect
3D sensors. The Kinect sensors produce streams of red, green and
blue (RGB) and depth images of 640 x 480 resolution at 30 frames per
second. Integrating the depth information to drivability gives a strong
cue to the analysis, which detects 3D results augmenting drivable and
non-drivable regions in 2D. The results of the 2D and 3D experiment
with different terrains, mines and tunnels, together with the qualitative
and quantitative evaluation, reveal that a good drivable region can be
detected in dynamic underground terrains.
Abstract: Automatic methods of detecting changes through
satellite imaging are the object of growing interest, especially
beca²use of numerous applications linked to analysis of the Earth’s
surface or the environment (monitoring vegetation, updating maps,
risk management, etc...). This work implemented spatial analysis
techniques by using images with different spatial and spectral
resolutions on different dates. The work was based on the principle
of control charts in order to set the upper and lower limits beyond
which a change would be noted. Later, the a contrario approach was
used. This was done by testing different thresholds for which the
difference calculated between two pixels was significant. Finally,
labeled images were considered, giving a particularly low difference
which meant that the number of “false changes” could be estimated
according to a given limit.
Abstract: Particle detection in very noisy and low contrast images
is an active field of research in image processing. In this article, a
method is proposed for the efficient detection and sizing of subsurface
spherical particles, which is used for the processing of softly fused
Au nanoparticles. Transmission Electron Microscopy is used for
imaging the nanoparticles, and the proposed algorithm has been
tested with the two-dimensional projected TEM images obtained.
Results are compared with the data obtained by transmission optical
spectroscopy, as well as with conventional circular object detection
algorithms.
Abstract: This paper describes a complex energy signal model
that is isomorphic with digital human fingerprint images. By using
signal models, the problem of fingerprint matching is transformed
into the signal processing problem of finding a correlation between
two complex signals that differ by phase-rotation and time-scaling. A
technique for minutiae matching that is independent of image
translation, rotation and linear-scaling, and is resistant to missing
minutiae is proposed. The method was tested using random data
points. The results show that for matching prints the scaling and
rotation angles are closely estimated and a stronger match will have a
higher correlation.
Abstract: Quality of 2D and 3D cross-sectional images produce
by Computed Tomography primarily depend upon the degree of
precision of primary and secondary X-Ray intensity detection.
Traditional method of primary intensity detection is apt to errors.
Recently the X-Ray intensity measurement system along with smart
X-Ray sensors is developed by our group which is able to detect
primary X-Ray intensity unerringly. In this study a new smart X-Ray
sensor is developed using Light-to-Frequency converter TSL230
from Texas Instruments which has numerous advantages in terms of
noiseless data acquisition and transmission. TSL230 construction is
based on a silicon photodiode which converts incoming X-Ray
radiation into the proportional current signal. A current to frequency
converter is attached to this photodiode on a single monolithic CMOS
integrated circuit which provides proportional frequency count to
incoming current signal in the form of the pulse train. The frequency
count is delivered to the center of PICDEM FS USB board with
PIC18F4550 microcontroller mounted on it. With highly compact
electronic hardware, this Demo Board efficiently read the smart
sensor output data. The frequency output approaches overcome
nonlinear behavior of sensors with analog output thus un-attenuated
X-Ray intensities could be measured precisely and better
normalization could be acquired in order to attain high resolution.
Abstract: Images of human iris contain specular highlights due
to the reflective properties of the cornea. This corneal reflection
causes many errors not only in iris and pupil center estimation but
also to locate iris and pupil boundaries especially for methods that
use active contour. Each iris recognition system has four steps:
Segmentation, Normalization, Encoding and Matching. In order to
address the corneal reflection, a novel reflection removal method is
proposed in this paper. Comparative experiments of two existing
methods for reflection removal method are evaluated on CASIA iris
image databases V3. The experimental results reveal that the
proposed algorithm provides higher performance in reflection
removal.
Abstract: Image Searching was always a problem specially when these images are not properly managed or these are distributed over different locations. Currently different techniques are used for image search. On one end, more features of the image are captured and stored to get better results. Storing and management of such features is itself a time consuming job. While on the other extreme if fewer features are stored the accuracy rate is not satisfactory. Same image stored with different visual properties can further reduce the rate of accuracy. In this paper we present a new concept of using polynomials of sorted histogram of the image. This approach need less overhead and can cope with the difference in visual features of image.
Abstract: In Multiple Sclerosis, pathological changes in the
brain results in deviations in signal intensity on Magnetic Resonance
Images (MRI). Quantitative analysis of these changes and their
correlation with clinical finding provides important information for
diagnosis. This constitutes the objective of our work. A new approach
is developed. After the enhancement of images contrast and the brain
extraction by mathematical morphology algorithm, we proceed to the
brain segmentation. Our approach is based on building statistical
model from data itself, for normal brain MRI and including clustering
tissue type. Then we detect signal abnormalities (MS lesions) as a
rejection class containing voxels that are not explained by the built
model. We validate the method on MR images of Multiple Sclerosis
patients by comparing its results with those of human expert
segmentation.
Abstract: This paper presents a system overview of Mobile to Server Face Recognition, which is a face recognition application developed specifically for mobile phones. Images taken from mobile phone cameras lack of quality due to the low resolution of the cameras. Thus, a prototype is developed to experiment the chosen method. However, this paper shows a result of system backbone without the face recognition functionality. The result demonstrated in this paper indicates that the interaction between mobile phones and server is successfully working. The result shown before the database is completely ready. The system testing is currently going on using real images and a mock-up database to test the functionality of the face recognition algorithm used in this system. An overview of the whole system including screenshots and system flow-chart are presented in this paper. This paper also presents the inspiration or motivation and the justification in developing this system.
Abstract: Retrieval image by shape similarity, given a template
shape is particularly challenging, owning to the difficulty to derive a
similarity measurement that closely conforms to the common
perception of similarity by humans. In this paper, a new method for the
representation and comparison of shapes is present which is based on
the shape matrix and snake model. It is scaling, rotation, translation
invariant. And it can retrieve the shape images with some missing or
occluded parts. In the method, the deformation spent by the template
to match the shape images and the matching degree is used to evaluate
the similarity between them.
Abstract: This paper describes an algorithm to estimate realtime vehicle velocity using image processing technique from the known camera calibration parameters. The presented algorithm involves several main steps. First, the moving object is extracted by utilizing frame differencing technique. Second, the object tracking method is applied and the speed is estimated based on the displacement of the object-s centroid. Several assumptions are listed to simplify the transformation of 2D images from 3D real-world images. The results obtained from the experiment have been compared to the estimated ground truth. From this experiment, it exhibits that the proposed algorithm has achieved the velocity accuracy estimation of about ± 1.7 km/h.
Abstract: Interpretation of aerial images is an important task in
various applications. Image segmentation can be viewed as the essential
step for extracting information from aerial images. Among many
developed segmentation methods, the technique of clustering has been
extensively investigated and used. However, determining the number
of clusters in an image is inherently a difficult problem, especially
when a priori information on the aerial image is unavailable. This
study proposes a support vector machine approach for clustering
aerial images. Three cluster validity indices, distance-based index,
Davies-Bouldin index, and Xie-Beni index, are utilized as quantitative
measures of the quality of clustering results. Comparisons on the
effectiveness of these indices and various parameters settings on the
proposed methods are conducted. Experimental results are provided
to illustrate the feasibility of the proposed approach.
Abstract: The number of features required to represent an image
can be very huge. Using all available features to recognize objects
can suffer from curse dimensionality. Feature selection and
extraction is the pre-processing step of image mining. Main issues in
analyzing images is the effective identification of features and
another one is extracting them. The mining problem that has been
focused is the grouping of features for different shapes. Experiments
have been conducted by using shape outline as the features. Shape
outline readings are put through normalization and dimensionality
reduction process using an eigenvector based method to produce a
new set of readings. After this pre-processing step data will be
grouped through their shapes. Through statistical analysis, these
readings together with peak measures a robust classification and
recognition process is achieved. Tests showed that the suggested
methods are able to automatically recognize objects through their
shapes. Finally, experiments also demonstrate the system invariance
to rotation, translation, scale, reflection and to a small degree of
distortion.
Abstract: In this paper, we present a comparative study between two computer vision systems for objects recognition and tracking, these algorithms describe two different approach based on regions constituted by a set of pixels which parameterized objects in shot sequences. For the image segmentation and objects detection, the FCM technique is used, the overlapping between cluster's distribution is minimized by the use of suitable color space (other that the RGB one). The first technique takes into account a priori probabilities governing the computation of various clusters to track objects. A Parzen kernel method is described and allows identifying the players in each frame, we also show the importance of standard deviation value research of the Gaussian probability density function. Region matching is carried out by an algorithm that operates on the Mahalanobis distance between region descriptors in two subsequent frames and uses singular value decomposition to compute a set of correspondences satisfying both the principle of proximity and the principle of exclusion.
Abstract: When reconstructing a scenario, it is necessary to
know the structure of the elements present on the scene to have an
interpretation. In this work we link 3D scenes reconstruction to
evolutionary algorithms through the vision stereo theory. We
consider vision stereo as a method that provides the reconstruction of
a scene using only a couple of images of the scene and performing
some computation. Through several images of a scene, captured from
different positions, vision stereo can give us an idea about the threedimensional
characteristics of the world. Vision stereo usually
requires of two cameras, making an analogy to the mammalian vision
system. In this work we employ only a camera, which is translated
along a path, capturing images every certain distance. As we can not
perform all computations required for an exhaustive reconstruction,
we employ an evolutionary algorithm to partially reconstruct the
scene in real time. The algorithm employed is the fly algorithm,
which employ “flies" to reconstruct the principal characteristics of
the world following certain evolutionary rules.
Abstract: This paper presented the potential of smart phone to
provide support on mapping the indoor asset. The advantage of using
the smart phone to generate the indoor map is that it has the ability to
capture, store and reproduces still or video images; indeed most of us
do have this powerful gadget. The captured images usually used by
maintenance team to save a record for future reference. Here, these
images are used to generate 3D models of an object precisely and
accurately for efficient and effective solution in data gathering. Thus,
it could be a resource for an informative database in asset
management.
Abstract: This frame work describes a computationally more
efficient and adaptive threshold estimation method for image
denoising in the wavelet domain based on Generalized Gaussian
Distribution (GGD) modeling of subband coefficients. In this
proposed method, the choice of the threshold estimation is carried out
by analysing the statistical parameters of the wavelet subband
coefficients like standard deviation, arithmetic mean and geometrical
mean. The noisy image is first decomposed into many levels to
obtain different frequency bands. Then soft thresholding method is
used to remove the noisy coefficients, by fixing the optimum
thresholding value by the proposed method. Experimental results on
several test images by using this method show that this method yields
significantly superior image quality and better Peak Signal to Noise
Ratio (PSNR). Here, to prove the efficiency of this method in image
denoising, we have compared this with various denoising methods
like wiener filter, Average filter, VisuShrink and BayesShrink.
Abstract: Freeways are originally designed to provide high
mobility to road users. However, the increase in population and
vehicle numbers has led to increasing congestions around the world.
Daily recurrent congestion substantially reduces the freeway capacity
when it is most needed. Building new highways and expanding the
existing ones is an expensive solution and impractical in many
situations. Intelligent and vision-based techniques can, however, be
efficient tools in monitoring highways and increasing the capacity of
the existing infrastructures. The crucial step for highway monitoring
is vehicle detection. In this paper, we propose one of such
techniques. The approach is based on artificial neural networks
(ANN) for vehicles detection and counting. The detection process
uses the freeway video images and starts by automatically extracting
the image background from the successive video frames. Once the
background is identified, subsequent frames are used to detect
moving objects through image subtraction. The result is segmented
using Sobel operator for edge detection. The ANN is, then, used in
the detection and counting phase. Applying this technique to the
busiest freeway in Riyadh (King Fahd Road) achieved higher than
98% detection accuracy despite the light intensity changes, the
occlusion situations, and shadows.