Search

A joint probabilistic face detection and tracking algorithm combining likelihood estimation and a prior probability is proposed in this paper. The likelihood estimation scheme is based on the statistical training of sets of automatically generated feature points and a mutual information tracking cue, while the prior probability estimation is based on a Gaussian temporal model. The likelihood estimation process is the core of a multiple face detection scheme used to initialize the tracking process. The resulting system was tested on real image sequences and is robust to significant partial occlusion and illumination changes. Index Terms--Arbitration scheme, illumination changes, mutual information, occlusion.

A theoretical performance analysis of watermarking schemes, based on correlation detection, is described and discussed.

A detection scheme is described for additive watermarking in the Discrete Cosine Transform (DCT) or Discrete Wavelet Transform (DWT) domains.

The L[subscript p] comparator is described and discussed. It can be used to improve performance in analog implementations of sorting networks.

In this paper, we propose an interpolation algorithm using a mathematical morphology morphing approach. The aim of this algorithm is to reconstruct the n-dimensional object from a group of (n -- 1)-dimensional sets representing sections of that object. The morphing transformation modifies pairs of consecutive sets such that they approach in shape and size. The interpolated set is achieved when the two consecutive sets are made idempotent by the morphing transformation. We prove the convergence of the morphological morphing. The entire object is modeled by successively interpolating a certain number of intermediary sets between each two consecutive given sets. We apply the interpolation algorithm for three-dimensional tooth reconstruction. Index Terms--Mathematical morphology, morphing, shape-based interpolation.

Research is presented concerning the use of the Bayesian theory to develop a method of predicting the location and velocity of a moving object. The use of algorithms, segmentation and optical flow tracking is discussed.

Research describing restoration methods for simulating the colors of old paintings are presented. Five digital color techniques requiring little computational expertise will allow use by art historians.

The authors examine the use of multichannel distance filters. Topics include time series vectors, digital filtering techniques, and motion fields.

By describing 3-D images as ellipsoids, the authors propose that they can utilize alpha-trimmed mean statistics to classify 3-D images. A discussion of the use of the Hough transform algorithm is presented.

An analog implementation of an order-statistics filter based on current-mode techniques is presented in this paper. The circuit is designed using switched-current delay lines and current maximum extractors. These filters could be easily incorporated to smart sensors as well as to smart cameras. SPICE simulation results demonstrate the feasibility of simple analog filters using current-mode techniques. Index Terms: Analog integrated circuits, analog signal processing, nonlinear filters.

The fuzzy scalar median is presented as a filter design methodology for digital signal and image processing. The proposed method uses ordering of fuzzy numbers based on fuzzy minimum and maximum operations defined by applying the extension principle and minimizing a fuzzy distance measure. It is shown that the equivalence of the two definitions is proven in experiments, while the basic properties of the fuzzy scalar median are investigated.

Watermark casting on digital images is an important problem since it affects many aspects of the information market. We propose a method for casting digital watermarks on images, and we analyze its effectiveness. The satisfaction of some basic demands in this area is examined, and a method for producing digital watermarks is proposed. Moreover, issues like immunity to subsampling and image-dependent watermarks are examined, and simulation results are provided for the verification of the above-mentioned topics. Index Terms - Copyright protecting, multimedia, watermark.

An analog implementation of running min/max filters based on current-mode techniques is presented in this brief. Switched-current delay cells and current/voltage two inputs min/max selectors are used either for current or voltage inputs respectively. The voltage two input Min/Max circuit is designed using current conveyors and a modified structure of this is used to implement the running Min/Max filter for window size n = 8. Simulation results demonstrate the feasibility of the proposed implementation, which can be extended to a higher window size. Index Terms - Min/Max filters, mixed analog-digital integrated circuits, nonlinear filters, running filters.

A study was conducted to evaluate the effectiveness of the median radial basis function (MRBF) neural network when used for estimating the optical flow and the segmentation of moving objects. Results show that the robust training algorithm employed by the MRBF neural network provides accurate and smooth approximations of optical flow and segmentation of motion. Furthermore, it was observed that only a minimal number of parameters were required by this approach to produce quality results.

The calculation of a class of running filters described as the n-ary extension of an associative, commutative, and idempotent binary operation T on an ordered sequence of operands was examined. The presence of a fast algorithm of complexity O(log(sub 2) n) T operations was proven for any arbitrary window filter of size n. A salient characteristic of the proof is its capability to produce a specific solution for every n.

The fuzzy set theory was introduced by Zadeh, to generalize the mathematical concept of the set to one of the fuzzy set. The use of fuzziness concepts in estimation theory is considered. The extension principle is used to fuzzify location and scale estimators. Fuzzy nonliner means are shown as extension of the corresponding crisp means, and the sample standard deviation sale estimator is extended to fuzzy numbers via the extension principle.

The least mean squares (LMS) based constrained and unconstrained adaptive L-filters suppress additive white noise encountered in image processing. Normalized, location-invariant and signed error LMS L-filters are compared and analyzed to minimize mean square error and mean absolute error between filter output and the noise-free signal. The normalized and the signed LMS L-filters have same efficiency for a given step-size parameter. A modified LMS L-filter that accelerates the rate of convergence and a signal-dependent adaptive filter that allows pixel treatment are given.

Morphological residual representations of one- and two-dimensional signals and image representations use constructive transforms to extract simple components for the representation of signals as a sum of their components. These representations give efficient encoding of the signal and describe the signal at varying levels. Experimental representations indicate that residual representations are effective in lossy image compression.

Radial basis functions (RBF's) consist of a two-layer neural network, where each hidden unit implements a kernel function. Each kernel is associated with an activation region from the input space and its output is fed to an output unit. In order to find the parameters of a neural network which embeds this structure we take into consideration two different statistical approaches. The first approach uses classical estimation in the learning stage and it is based on the learning vector quantization algorithm and its second-order statistics extension. After the presentation of this approach, we introduce the median radial basis function (MRBF) algorithm based on robust estimation of the hidden unit parameters. The proposed algorithm employs the marginal median for kernel location estimation and the median of the absolute deviations for the scale parameter estimation. A histogram-based fast implementation is provided for the MRBF algorithm. The theoretical performance of the two training algorithms is comparatively evaluated when estimating the network weights. The network is applied in pattern classification problems and in optical flow segmentation.

Orthogonal matrix functions and orthogonal transforms are introduced into a Fourier analysis of multichannel time series used in signal and image processing. The novel approach to multichannel time series Fourier analysis offers characteristics similar to classical Fourier series. Consequently, it serves as a viable general framework for the definition of other multichannel transforms.

Parallel algorithms on barrel shifter computers for one- and two-dimensional signal processing operators have a running maximum/minimum and moving average filter, and running sorting and sliding window fast Fourier transforms. Their implementation on reduced barrel shifter computer networks reveal that they are in consent with these networks. The extension of two dimensional algorithms implements sequences of length larger than the number of processing elements.

The design of a class of learning vector quantizers (LVQ) involves the use multivariate order statistics that shows robust properties. The median LVQ is a special class of LVQ that uses either the marginal median or the vector median as a multivariate estimator of location. The performance of marginal median LVQ in color image quantization, in the presence of noise, is superior to that of classical LVQ. The execution time of the algorithm is also short.

Multichannel AR modeling and multichannel histogram equalization are used for color imaging and color image equalization, respectively. A new method is proposed for the multichannel equalization that equalizes the n-dimensional histogram of a n-channel image. Color image localization is performed simultaneously on the three channels RGB, using the joint pdf. Implementation of a parallel algorithm leads to the calculation and equalization of color image histogram. Multichannel AR models are used for segmenting and synthesizing color texture regions.

The subordering principle of marginal data ordering is used for extending single channel L filters to multichannel case. Mean squared error (MSE) is required as a fidelity criterion for using the marginal data ordering and design of multichannel L filters. Subordering enables the description of a vector quantization-based unified framework for discrete calculations. Extended cases are used to filter artificially generated velocity fields.

A new method which incorporated the simulated annealing algorithms, was used to choose optimal discrete-valued digital filter parameters. The algorithms were more efficient than iterative improvement algorithms when applied to median filter length adaptations. The simulated annealing algorithms optimized linear filter designs.

This paper deals with the design of multichannel Wiener filters both in the spatial and frequency domain. FIR and IIR Wiener filters are presented. Color image restoration based on multi-channel autoregressive (AR) image modeling is examined. Detailed discussions on the use of a multichannel Wiener filter in color image restoration incorporating the interchannel correlations and computer simulations are presented also.

Ultrasonic image segmentation, when combined with filtering, results in better accurate lesion detection and edge preservation and a high ultrasonic speckle reduction without the loss of image detail. Segmentation of ultrasonic images are obtained using self organized neural networks. Signal-adaptive filters are used in ultrasonic speckle suppression.

A process is presented for three-dimensional shape registration based on the iterative closest point algorithm, and enhanced via a 3-D volume which contains the shapes to be registered.

In this paper, the morphological skeleton interpolation (MSI) algorithm is presented. It is an efficient, shape-based interpolation method used for interpolating slices in a three-dimensional (3-D) binary object. It is based on morphological skeletonization, which is used for two-dimensional (2-D) slice representation. The proposed morphological skeleton matching process provides translation, rotation, and scaling information at the same time. The interpolated slices preserve the shape of the original object slices, when the slices have similar shapes. It can also modify the shape of an object when the successive slices do not have similar shapes. Applications on artificial and real data are also presented. Index Terms--Mathematical morphology, shape representation, skeletonization, three-dimensional (3-D) binary image interpolation.

The MPEG-2 compression algorithm is very sensitive to channel disturbances due to the use of variable-length coding. A single bit error during transmission leads to noticeable degradation of the decoded sequence quality, in that part or an entire slice information is lost until the next resynchronization point is reached. Error concealment (EC) methods, implemented at the decoder side, present one way of dealing with this problem. An error-concealment scheme that is based on block-matching principles and spatio-temporal video redundancy is presented in this paper. Spatial information (for the first frame of the sequence or the next scene) or temporal information (for the other frames) is used to re. construct the corrupted regions. The concealment strategy is embedded in the MPEG-2 decoder model in such a way that error concealment is applied after entire frame decoding. Its performance proves to be satisfactory for packet error rates (PER) ranging from 1% to 10% and for video sequences with different content and motion and surpasses that of other EC methods under study. Index Terms--Block matching, error concealment, MPEG-2 compression.

A generalized fuzzy mathematical morphology is applied to robust 2-D and 3-D object representation.

A study was conducted to analyze edge detection on angular signals using filters. A series of one-dimensional angular signal experiments were performed to describe the operation of the filters and edge detectors. Several properties were also obtained for the circular media filter. Results indicated that circular quasiranges can be utilized to successfully determine edge detectors on hue images.

This paper deals with the statistical analysis of the behavior of a blind robust watermarking system based on pseudorandom signals embedded in the magnitude of the Fourier transform of the host data. The host data that the watermark is embedded into is one-dimensional and non-white, following a specific probability model. The analysis performed involves theoretical evaluation of the statistics of the Fourier coefficients and the design of an optimal detector for multiplicative watermark embedding. Finally, experimental results are presented in order to show the performance of the proposed detector versus that of the correlator detector.

Visual speech recognition is an emerging research field. In this paper, we examine the suitability of support vector machines for visual speech recognition. Each word is modeled as a temporal sequence of visemes corresponding to the different phones realized. One support vector machine is trained to recognize each viseme and its output is converted to a posterior probability through a sigmoidal mapping. To model the temporal character of speech, the support vector machines are integrated as nodes into a Viterbi lattice. We test the performance of the proposed approach on a small visual speech recognition task, namely the recognition of the first four digits in English. The word recognition rate obtained is at the level of the previous best reported rates.