Abstract: This thesis studies rapid object detection, focusing
on feature-based methods. Firstly, modifications of training and
detection of the Viola-Jones method are made to improve
performance and overcome some of the current limitations such as
rotation, occlusion and articulation. New classifiers produced by
training and by converting existing classifiers are tested in face
detection and hand detection. Secondly, the nature of invariant
features in terms of the computational complexity, discrimination
power and invariance to rotation and scaling are discussed. A new
feature extraction method called Concentric Discs Moment Invariants
(CDMI) is developed based on moment invariants and summed-area
tables. The dimensionality of this set of features can be increased
by using additional concentric discs, rather than using higher order
moments. The CDMI set has useful properties, such as speed, rotation
invariance, scaling invariance, and rapid contrast stretching can be
easily implemented. The results of experiments with face detection
shows a clear improvement in accuracy and performance of the CDMI
method compared to the standard moment invariants method. Both
the CDMI and its variant, using central moments from concentric
squares, are used to assess the strength of the method applied to
hand-written digits recognition. Finally, the parallelisation of the
detection algorithm is discussed. A new model for the specific case
of the Viola-Jones method is proposed and tested experimentally.
This model takes advantage of the structure of classifiers and of the
multi-resolution approach associated with the detection method. The
model shows that high speedups can be achieved by broadcasting
frames and carrying out the computation of one or more cascades in
each node.
Key Words: Human Face Recognition, Pattern Recognition
Systems, Computer Vision
Barczak, A.L.C., “Comparação entre algoritmos de mínimos quadrados e de
zona mínima para desvios de circularidade (A Comparison BetweenLeast-Square Center Algorithms and Minimum Zone Center Algorithms ForCircularity Assessment”, Master’s Thesis, Unicamp, Brazil, 1996.
Abstract: The precise assessment of out-of-roundness is an
important issue in metrology. There are always errors associated
with these measurements, caused by different sources. The computer
aided measuring machines have an additional source of error, the
algorithm itself. This subject was only considered important in recent
times. The main objective in this work is to determine the behavior
of minimum zone center (MZC), conceptually the most correct, in
comparison with the least square center (LSC), used in practice. In
order to analyze the differences between the algorithms’ answers,
computer simulations were used. Both algorithms were developed in
C language, and a third program generated thousands of sets similar
to those obtained through CMMs. The algorithm using Voronoi
diagrams (MZC type) was developed taking into consideration the
optimization assurance, low complexity and implementation facility.
The out-of- roundness values, numbers of points of the set and the
radius of the workpiece simulated an industrial environment. The
results showed that the Voronoi algorithm gets a pair of concentric
circles with separation equal or less then the LSC one. The difference
between the results could be about 30% or greater, depending
on the points’ distribution. The behavior of the maximum errors
between results depends on the number of points and the value
interval of out-of-roundness. For some CMMs used in industry, the
uncertainty of the machine is greater then the maximum error caused
by algorithm’s choice. This is true for various number of points
and out-of-roundness values. In these cases, the LSC is accurate
enough. Also when only LSC is available, the maximum expected
error could be minimized by increasing the number of points. For
dedicated machines and CMMs that are more accurate, the algorithm
is an important source of error. It is always necessary to use MZC
algorithms in these cases. In view of these results, one presumes
that the development of other minimum zone algorithms, such as
cilindricity and sphericity, will be very important in future. (NOTE:
This is an updated version of the abstract available in the above
URL).
The results of this work were published in a peer-reviewed journal 5, cited 10
times (in 2008(2), 2007, 2006(2), 2004, 2003, 2002, 2000(2)) according to ISI
Web of Knowledge.