Emmanouil Z. Psarakis

Parametric Image Alignment Using Enhanced Correlation Coefficient Maximization

In this work we propose the use of a modified version of the correlation coefficient as a performance criterion for the image alignment problem. The proposed modification has the desirable characteristic of being invariant with respect to photometric distortions. Since the resulting similarity measure is a nonlinear function of the warp parameters, we develop two iterative schemes for its maximization, one based on the forward additive approach and the second on the inverse compositional method. As it is customary in iterative optimization, in each iteration the nonlinear objective function is approximated by an alternative expression for which the corresponding optimization is simple. In our case we propose an efficient approximation that leads to a closed form solution (per iteration) which is of low computational complexity, the latter property being particularly strong in our inverse version. The proposed schemes are tested against the forward additive Lucas-Kanade and the simultaneous inverse compositional algorithm through simulations. Under noisy conditions and photometric distortions our forward version achieves more accurate alignments and exhibits faster convergence whereas our inverse version has similar performance as the simultaneous inverse compositional algorithm but at a lower computational complexity.

Design of two-dimensional zero phase FIR fan filters via the McClellan transform

The authors present a method for the design of 2-D zero-phase finite-impulse-response (FIR) fan filters with quadrantal symmetry using the McClellan transform. They give conditions that the coefficients of the McClellan transform must satisfy in order to avoid the scaling of the transform. The proposed design method satisfies these conditions. The resulting cut-off isopotential of the design method is shown to have a very small relative absolute deviation from the ideal one. The method is extended to the design of 2-D zero-phase FIR fan filters of general shape. >

An enhanced correlation-based method for stereo correspondence with subpixel accuracy

The invariance of the similarity measure in photometric distortions as well as its capability in producing subpixel accuracy are two desired and often required features in most stereo vision applications. In this paper we propose a new correlation-based measure which incorporates both mentioned requirements. Specifically, by using an appropriate interpolation scheme in the candidate windows of the matching image, and using the classical zero mean normalized cross correlation function, we introduce a suitable measure. Although the proposed measure is a nonlinear function of the sub-pixel displacement parameter, its maximization results in a closed form solution, resulting in reduced complexity for its use in matching techniques. Application of the proposed measure in a number of benchmark stereo pair images reveals its superiority over existing correlation-based techniques used for subpixel accuracy.

Design of two-dimensional zero-phase FIR filters via the generalized McClellan transform

The authors present a design method for two-dimensional centrosymmetric zero-phase finite-impulse-response (FIR) filters, via the generalized McClellan transform. An in-depth study of the transform involved in the design method reveals a number of useful properties. These properties are used in the design method for an optimal definition of the generalized McClellan transform coefficients. The method can be applied to the design of classical 2-D FIR filters, yielding filters that are very close to the 2-D ideal specifications. >

An L/sub 2/-based method for the design of 1-D zero phase FIR digital filters

Finite impulse response (FIR) filters obtained with the classical L/sub 2/ method have performance that is very sensitive to the form of the ideal response selected for the transition region. It is known that design requirements do not constrain in any way the ideal response inside this region. Most existing techniques utilize this flexibility. By selecting various classes of functions to describe the undefined part of the ideal response they develop methods that improve the performance of the L/sub 2/ based filters. In this paper we propose a means for selecting the unknown part of the ideal response optimally. Specifically by using a well-known property of the Fourier approximation theory we introduce a suitable quality measure. The proposed measure is a functional of the ideal response and depends on its actual form inside the transition region. Using variational techniques we succeed in minimizing the introduced criterion with respect to the ideal response and thus obtain its corresponding optimum form. The complete solution to the problem can be obtained by solving a simple system of linear equations suggesting a reduced complexity for the proposed method. An extensive number of design examples show the definite superiority of our method over most existing non min-max design techniques, while the method compares very favorably with min-max optimum methods. Finally we prove that the approximation error function of our filter has the right number of alternating extrema, required by the L/sub /spl infin// criterion, in the passband and stopband. This results in a significant convergence speed up, if our optimum solution is used as an initialization scheme, of the Remez exchange algorithm.

A Generative Model for the Joint Registration of Multiple Point Sets

This paper describes a probabilistic generative model and its associated algorithm to jointly register multiple point sets. The vast majority of state-of-the-art registration techniques select one of the sets as the “model” and perform pairwise alignments between the other sets and this set. The main drawback of this mode of operation is that there is no guarantee that the model-set is free of noise and outliers, which contaminates the estimation of the registration parameters. Unlike previous work, the proposed method treats all the point sets on an equal footing: they are realizations of a Gaussian mixture (GMM) and the registration is cast into a clustering problem. We formally derive an EM algorithm that estimates both the GMM parameters and the rotations and translations that map each individual set onto the “central” model. The mixture means play the role of the registered set of points while the variances provide rich information about the quality of the registration. We thoroughly validate the proposed method with challenging datasets, we compare it with several state-of-the-art methods, and we show its potential for fusing real depth data.

An efficient spatial domain technique for subpixel image registration

In this paper a new technique for performing image registration with subpixel accuracy is presented. The proposed technique, which is based on the maximization of the correlation coefficient function, does not require the reconstruction of the intensity values and provides a closed-form solution to the subpixel translation estimation problem. Moreover, an efficient iterative scheme is proposed, which reduces considerably the overall computational cost of the image registration problem. This scheme properly combined with the proposed similarity measure results in a fast spatial domain technique for subpixel image registration.

Computer Aided Detection of Prostate Cancer using Fused Information from Dynamic Contrast Enhanced and Morphological Magnetic Resonance Images

This paper presents a computer-aided diagnosis scheme for the detection of prostate cancer. The pattern recognition scheme proposed, utilizes fused dynamic and morphological features extracted from magnetic resonance images (MRIs). The performance of the proposed scheme has been evaluated through extensive training and testing on several patient cases, where the staging of their condition has been previously evaluated by both ultrasoundguided biopsy and radiological assessment. The classification scheme is based on Probabilistic Neural Networks (PNNs), whose parameters are estimated using the Expectation-Maximization (EM) algorithm during a training phase. Fusion of the image characteristics is performed by properly aligning the respective T1-weighted dynamic and T2-weighted morphological images, allowing accurate feature selection from both images. The proposed classification scheme as well as the effect of fusion on the extracted features is tested, with respect to the correct classification rate (CCR) of each case.

Automatic Kronecker Product Model Based Detection of Repeated Patterns in 2D Urban Images

Repeated patterns (such as windows, tiles, balconies and doors) are prominent and significant features in urban scenes. Therefore, detection of these repeated patterns becomes very important for city scene analysis. This paper attacks the problem of repeated patterns detection in a precise, efficient and automatic way, by combining traditional feature extraction followed by a Kronecker product low-rank modeling approach. Our method is tailored for 2D images of building facades. We have developed algorithms for automatic selection of a representative texture within facade images using vanishing points and Harris corners. After rectifying the input images, we describe novel algorithms that extract repeated patterns by using Kronecker product based modeling that is based on a solid theoretical foundation. Our approach is unique and has not ever been used for facade analysis. We have tested our algorithms in a large set of images.

A computer-aided system for the detection of prostate cancer based on magnetic resonance image analysis

This paper presents an overview of a computer- aided system for the detection of carcinomas in the prostate gland. The proposed system incorporates information from two different types of magnetic resonance images (MRIs), namely the T2-weighted morphological images and the T1-weighted dynamic contrast enhanced (DCE) images, to extract discriminative features that will be used in the training phase of a classification algorithm for the differentiation between malignant and benign tissue. The resulting feature vectors are also used for the assessment of new patient cases. The pattern recognition scheme is based on probabilistic neural networks (PNNs). The parameters of the PNNs are estimated using the expectation- maximization (EM) algorithm. The performance of the proposed computer-aided detection system is evaluated through training and testing on several patient cases, whose condition has been previously assessed through ultrasound-guided biopsy and MRI examination.