We have seen an explosion of data exchange in the Internet and the extensive use of digital media. Consequently, digital data owners can quickly and massively transfer multimedia documents across the Internet. This leads to wide interest in multimedia security and multimedia copyright protection. We propose a novel hybrid digital video watermarking scheme based on the scene change analysis and error correction code. Our video watermarking algorithm is robust against the attacks of frame dropping, averaging and statistical analysis, which were not solved effectively in the past. We start with a complete survey of current watermarking technologies, and noticed that none of the existing schemes is capable of resisting all attacks. Accordingly, we propose the idea of embedding different parts of a single watermark into different scenes of a video. We then analyze the strengths of different watermarking schemes, and apply a hybrid approach to form a super watermarking scheme that can resist most of the attacks. To increase the robustness of the scheme, the watermark is refined by an error correcting code, while the correcting code is embedded as a watermark in the audio channel. It optimizes the quality of the watermarked video. The effectiveness of this scheme is verified through a series of experiments, in which a number of standard image processing attacks are conducted, and the robustness of our approach is demonstrated using the criteria of the latest StirMark test.

To achieve integrated segmentation and recognition in complex scenes, the model-based approach has widely been accepted as a promising paradigm. However, the performance is still far from satisfactory when the target object is highly deformed and the level of outlier contamination is high. In this paper, we first describe two Bayesian frameworks, one for classifying input patterns and another for detecting target patterns in complex scenes using deformable models. Then, we show that the two frameworks are similar to the forward-reverse setting of Hausdorff matching and that their matching and discriminating properties are complementary to each other. By properly combining the two frameworks, we propose a new matching scheme called bidirectional matching. This combined approach inherits the advantages of the two Bayesian frameworks. In particular, we have obtained encouraging empirical results on shape-based pattern extraction, using a subset of the CEDAR handwriting database containing handwritten words of highly varying shape.

We present a novel approach for video partitioning by detecting three essential types of camera breaks, namely cuts, wipes, and dissolves. The approach is based on the analysis of temporal slices which are extracted from the video by slicing through the sequence of video frames and collecting temporal signatures. Each of these slices contains both spatial and temporal information from which coherent regions are indicative of uninterrupted video partitions separated by camera breaks. Properties could further be extracted from the slice for both the detection and classification of camera breaks. For example, cut and wipes are detected by color-texture properties, while dissolves are detected by statistical characteristics. The approach has been tested by extensive experiments.

We present a scheme for automatically partitioning videos into scenes. A scene is generally referred to as a group of shots taken at the same site. We first propose a motion annotation algorithm based on the analysis of spatiotemporal image volumes. The algorithm characterizes the motions within shots by extracting and analyzing the motion trajectories encoded in the temporal slices of image volumes. A motion-based keyframe computing and selection strategy is thus proposed to compactly represent the content of shots. With these techniques, we further present a scene change detection algorithm by measuring the similarity of the representative keyframes in shots.

In this paper, we present approaches for detecting camera cuts, wipes and dissolves based on the analysis of spatio-temporal slices obtained from videos. These slices are composed of spatially and temporally coherent regions which can be perceived as shots. In the proposed methods, camera breaks are located by performing color-texture segmentation and statistical analysis on these video slices. In addition to detecting camera breaks, our methods can classify the detected breaks as camera cuts, wipes and dissolves in an efficient manner.

A Bayesian framework for deformable pattern classification was proposed by K.W. Cheung et al. (1998), with promising results for isolated handwritten character recognition. Its performance, however degrades significantly when it is applied to detect deformable patterns in complex scenes, where the amount of outliers due to other neighboring objects or the background is usually large. Also, the fact that the associated evidence measure does not penalize models resting on white space results in a high false alarm rate. Another Bayesian framework for deformable pattern detection is proposed. The framework possesses the intrinsic property of matching with only part of an image (segmentation) and its associated evidence measure can penalize white space implicitly. However, limited data exploration capability is the major trade-off. By properly combining the two frameworks, a new matching algorithm called bidirectional matching is proposed. This combined approach possesses the advantages of the two frameworks and gives robust results for non-rigid shape extraction. To evaluate the performance of the proposed approach, we have applied it to shape-based handwritten word retrieval. Using a subset of the bb dataset in the CEDAR database, we can achieve a recall rate of 59% and a precision rate of 43%.

Deformable models have recently been proposed for many pattern recognition applications due to their ability to handle large shape variations. These proposed approaches represent patterns or shapes as deformable models, which deform themselves to match with the input image, and subsequently feed the extracted information into a classifier. The three components-modeling, matching, and classification-are often treated as independent tasks. In this paper, we study how to integrate deformable models into a Bayesian framework as a unified approach for modeling, matching, and classifying shapes. Handwritten character recognition serves as a testbed for evaluating the approach. With the use of our system, recognition is invariant to affine transformation as well as other handwriting variations. In addition, no preprocessing or manual setting of hyperparameters (e.g., regularization parameter and character width) is required. Besides, issues on the incorporation of constraints on model flexibility, detection of subparts, and speed-up are investigated. Using a model set with only 23 prototypes without any discriminative training, we can achieve an accuracy of 94.7 percent with no rejection on a subset (11,791 images by 100 writers) of handwritten digits from the NIST SD-1 dataset.

Summary form only given. It is a well recognized fact that economic growth in industrial countries come from three main sources-technical progress, capital, and a well-educated labour force. To maintain our position as, or to become, economic powerhouses in the world, education in science and technology must be emphasized and wide spread. Computer education (not just software education) is an important center piece. To train engineers and scientists for the future, we should not narrowly focus our attention just to software education. The traditional computer science curriculum is no longer adequate for today's needs. Most problems and applications today require integrated solutions involving hardware and software. Our education system should be designed to provide students with a broad background of fundamentals and system approach training involving hardware, software, networking, and communications. I challenge this panel to come up with a sketch of such a curriculum on "software" education for Asia-Pacific in the 21st century.

Following the success of applying deformable models to feature extraction, a natural next step is to apply such models to pattern classification. Recently, we have cast a deformable model under a Bayesian framework for classification, giving promising results. However, deformable model methods are computationally expensive due to the required iterative optimization process. The problem is even more severe when there are a large number of models (e.g., for character recognition), because each of them has to deform and match with the input data before a final classification can be derived. In this paper, we propose to combine the deformable models into a mixture, in which the individual models compete with each other to survive the matching process during classification. Models that do not compete well are eliminated early, thus allowing substantial savings in computation. This process of competition-elimination has been applied to handwritten digit recognition in which significant speedup can be achieved without sacrificing recognition accuracy.

The authors comments that Charif-Chefchaouni and Schonfeld (see ibid., vol.3, no.6, p.847, 1994) investigated the invertibility of a morphological representation of binary images and determined the necessary and sufficient conditions for its inverse. The authors show that one of the derived necessary conditions is not valid. A counterexample is given to illustrate our observations. Charif-Chefchaouni and Schonfeld reply that the new sufficient condition is proposed for the invertibility of the morphological image representation. A modification of its inverse is subsequently used to derive a new necessary condition for the invertibility of the morphological image representation. A composition of these conditions is finally used to provide a new necessary and sufficient condition under some restrictions for the invertibility of the morphological image representation. These necessary and sufficient conditions form a revision of one of the necessary conditions for the invertibility of the morphological image representation stated in the original paper.

The number of nerve fibers (axons) in a nerve, the axon size, and shape can all be important neuroanatomical features in understanding different aspects of nerves in the brain. However, the number of axons in a nerve is typically in the order of tens of thousands and a study of a particular aspect of the nerve often involves many nerves. Potentially meaningful studies are often prohibited by the huge number involved when manual measurements have to be employed. A method that automates the analysis of axons from electron-micrographic images is presented. It begins with a rough identification of all the axon centers by use of an elliptical Hough transform procedure. Boundaries of each axons are then extracted based on active contour model, or snakes, approach where physical properties of the axons and the given image data are used in an optimization scheme to guide the snakes to converge to axon boundaries for accurate sheath measurement. However, false axon detection is still common due to poor image quality and the presence of other irrelevant cell features, thus a conflict resolution scheme is developed to eliminate false axons to further improve the performance of detection. The developed method has been tested on a number of nerve images and its results are presented.

This paper considers the problem of modeling and extracting arbitrary deformable contours from noisy images. We propose a global contour model based on a stable and regenerative shape matrix, which is invariant and unique under rigid motions. Combined with Markov random field to model local deformations, this yields prior distribution that exerts influence over a global model while allowing for deformations. We then cast the problem of extraction into posterior estimation and show its equivalence to energy minimization of a generalized active contour model. We discuss pertinent issues in shape training, energy minimization, line search strategies, minimax regularization and initialization by generalized Hough transform. Finally, we present experimental results and compare its performance to rigid template matching.>

For image processing systems that have a limited size of region of support, say 3/spl times/3, direct implementation of morphological operations by a structuring element larger than the prefixed size is impossible. The decomposition of morphological operations by a large structuring element into a sequence of recursive operations, each using a smaller structuring element, enables the implementation of large morphological operations. In this paper, the authors present the decomposition of arbitrarily shaped (convex or concave) structuring elements into 3/spl times/3 elements, optimized with respect to the number of 3/spl times/3 elements. The decomposition is based on the concept of factorization of a structuring element into its prime factors. For a given structuring element, all its corresponding 3/spl times/3 prime concave factors are first determined. From the set of the prime factors, the decomposability of the structuring element is then established, and subsequently the structuring element is decomposed into a smallest possible set of 3/spl times/3 elements. Examples of optimal decomposition and structuring elements that are not decomposable are presented.>

This paper considers the problem of modeling and extracting arbitrary deformable contours from noisy images. We propose a global contour model based on a stable and regenerative shape matrix, which is invariant and unique under rigid motions. Combined with Markov random field to model local deformations, this yields prior distribution that exerts influence over a global model while allowing for deformations. We then cast the problem of extraction into posterior estimation and show its equivalence to energy minimization of a generalized active contour model. We discuss pertinent issues in shape training, minimax regularization and initialization by generalized Hough transform. Finally, we present experimental results and compare its performance to rigid template matching.>

We describe a multiple-scale boundary representation based on morphological operations. An object boundary is first progressively smoothed by a number of opening and closing operations using a set of structuring elements of increasing size, generating a multiple scale representation of the object. Then, corresponding features of the smoothed boundary across a continuum of scales are extracted and linked together forming a map called the morphological scale space. Properties of this scale space map are investigated and presented. A shape smoothing algorithm, which is based on this map and its properties, is proposed to show how the scale space representation could be applied to solve practical image processing problems. Specifically, in line with Witkin's (1983) Gaussian scale space filtering, boundary features that are explicitly related across scales by the morphological scale space map are organized into global region information and local boundary features. From the organization, perceptually dominant features are determined and a shape is smoothed without the requirement of prior knowledge of the object boundary nor its noise.>

A morphological operation using a large structuring element can be decomposed equivalently into a sequence of recursive operations, each using a smaller structuring element. However, an optimal decomposition of arbitrarily shaped structuring elements is yet to be found. In this paper, we have derived an optimal decomposition of a specific class of structuring elements/spl mdash/convex sets/spl mdash/for a specific type of machine/spl mdash/4-connected parallel array processors. The cost of morphological operation on 4-connected parallel array processors is the total number of 4-connected shifts required by the set of structuring elements. First, the original structuring element is decomposed into a set of prime factors, and then their locations are determined while minimizing the cost function. Proofs are presented to show the optimality of the decomposition. Examples of optimal decomposition are given and compared to an existing decomposition reported by Xu (1991).>

This paper describes a reconstructable thinning process which is based on one-pass parallel thinning and the morphological skeleton transformation. It reduces a binary digital pattern into a unit-width connected skeleton enabling perfect reconstruction of the original pattern. The process uses thinning templates to iteratively remove boundary pixels and structuring templates of the morphological skeleton transformation to retain critical feature pixels for reconstruction. The thinning templates together with the extracted feature pixels ensure skeletal connectivity, unit width, and reconstructability. These essential properties are guaranteed regardless of the chosen structuring templates used in the morphological skeleton transformation. The thinning process is analyzed and results are presented. A number of implementation issues, such as the choice of structuring templates, the computational model, noise filtering, and computational efficiency, are also addressed.

Investigates the estimation of 2-D boundary functions from sampled data sets where both noise and corners are present. The approach is based on the partial smoothing spline in which the estimated boundary function consists of an ordinary smoothing spline and a parametric function that describes the discontinuities (i.e., corners of the boundary). Prior knowledge about the boundary, such as the number of corners, their locations, noise levels, and the amount of smoothness, is not required for the boundary estimate. The smoothing parameter and the corner locations of the spline, which are parts of the estimate, are determined by the generalized cross-validation method whereby statistical properties are gathered from the input sampled data rather than specified a priori. This approach enables the smoothing of a noisy boundary while retaining an accurate description of the boundary corners. Extensive experiments were conducted to verify its ability to smooth noise while retaining a good representation of boundary corners, and do not rely on any prior information.>

A one-pass parallel thinning algorithm based on a number of criteria, including connectivity, unit-width convergence, medial axis approximation, noise immunity, and efficiency, is proposed. A pipeline processing model is assumed for the development. Precise analysis of the thinning process is presented to show its properties, and proofs of skeletal connectivity and convergence are provided. The proposed algorithm is further extended to the derived-grid to attain an isotropic medial axis representation. A set of measures based on the desired properties of thinning is used for quantitative evaluation of various algorithms. Image reconstruction from connected skeletons is also discussed. Evaluation shows that the procedures compare favorably to others.>

A technique for detecting and localizing corners of planar curves is proposed. The technique is based on Gaussian scale space, which consists of the maxima of absolute curvature of the boundary function presented at all scales. The scale space of isolated simple and double corners is first analyzed to investigate the behavior of scale space due to smoothing and interactions between two adjacent corners. The analysis shows that the resulting scale space contains line patterns that either persist, terminate, or merge with a neighboring line. Next, the scale space is transformed into a tree that provides simple but concise representation of corners at multiple scales. Finally, a multiple-scale corner detection scheme is developed using a coarse-to-fine tree parsing technique. The parsing scheme is based on a stability criterion that states that the presence of a corner must concur with a curvature maximum observable at a majority of scales. Experiments were performed to show that the scale space corner detector is reliable for objects with multiple-size features and noisy boundaries and compares favorably with other corner detectors tested.>

The problem of multi-channel restoration using both within and between-channel deterministic information is considered. A multi-channel image is a set of image planes that exhibit cross-plane similarity. Existing optimal restoration filters for single-plane images will yield suboptimal results when applied to multi-channel images, since between-channel information is not utilized. Multi-channel least squares restoration filters are developed using two approaches, the set theoretic and the constrained optimization. A geometric interpretation of the estimates of both filters is given. Color images, that is three-channel imagery with red, green, and blue components, are considered. Constraints that capture the within and between-channel properties of color images are developed. Issues associated with the computation of the two estimates are addressed. Finally, experiments using color images are shown.>

Wiener filtering, an iterative procedure which successively uses the Wiener-filtered signal as an improved prototype to update the covariance estimates, is investigated. The convergent properties of this iterative Wiener filter are analyzed. It has been shown that the iterative Wiener filter converges to a fixed point which does not correspond to the true covariance. Based on the analysis presented, and iterative filter is proposed to correct for the convergence error which results in suboptimal performance of the original iterative Wiener filter. The performance of this filter is shown to be theoretically optimal. Experiments are conducted in a practical setting to demonstrate the effectiveness of the proposed methods.>

A technique for detecting and localizing corners of planar curves is proposed. The technique is based on Gaussian scale space, which consists of the maxima of absolute curvature of the boundary function presented at all scales. The scale space of isolated simple and double corners is analyzed to investigate the behavior of scale space due to smoothing and interactions between two adjacent corners. The scale space is transformed into a tree which provides simple but concise representation of corners at multiple scales. A multiple-scale corner detection scheme is developed using a coarse-to-fine tree parsing technique. The parsing scheme is based on a stability criterion which states that the presence of a corner must concur with a curvature maximum observable at a majority of scales. Experimental results show that the scale-space corner detector is reliable for objects with multiple-size features and noisy boundaries and that it compares favorably with other corner detectors tested.>

A precise definition of digital skeletons and a mathematical framework for the analysis of a class of thinning algorithms, based on morphological set transformation, are presented. A particular thinning algorithm (algorithm A) is used as an example in the analysis. Precise definitions and analyses associated with the thinning process are presented, including the proof of convergence, the condition for one-pixel-thick skeletons, and the connectedness of skeletons. In addition, a necessary and sufficient condition for the thinning process in general is derived, and an algorithm (algorithm B) based on this condition is developed. Experimental results are used to compare the two thinning algorithms, and issues involving noise immunity and skeletal bias are addressed.>

A parallel algorithm is presented for detecting dominant points on a digital closed curve. The procedure requires no input parameter and remains reliable even when features of multiple sizes are present on the digital curve. The procedure first determines the region of support for each point based on its local properties, then computes measures of relative significance (e.g. curvature) of each point, and finally detects dominant points by a process of nonmaximum suppression. This procedure leads to the observation that the performance of dominant points detection depends not only on the accuracy of the measure of significance, but also on the precise determination of the region of support. This solves the fundamental problem of scale factor selection encountered in various dominant point detection algorithms. The inherent nature of scale-space filtering in the procedure is addressed, and the performance of the procedure is compared to those of several other dominant point detection algorithms, using a number of examples.>

A method is presented for estimating the orientation of a textured surface as a descriptor of surface shape. It is based on the analysis of local spectrum information of the texture in an image. The local spectrum representation is computed by the two-dimensional Wigner distribution, which gives the spatial-frequency information as a function of location. The change in texture density, or the so-called texture gradient, caused by the perspective projection of a surface in the three dimensional world onto the two-dimensional image plane, is computed from this space-frequency representation by measuring the high-frequency energy distribution at each location of the image. The surface orientation is then estimated from the texture gradient. This method was implemented for the limited case of planar surfaces. Simulations were performed and results were analyzed to address issues related to the method's estimation accuracy, implementation, and limitations.>

A parallel algorithm for detecting dominant points on a digital closed curve is presented. The procedure requires no input parameter and remains reliable even when features of multiple sizes are present on the digital curve. The procedure first determines the region of support for each point based on its local properties, then computes measures of relative significance (e.g. curvature) of each point, and finally detects dominant points by a process of nonmaxima suppression. This procedure leads to an important observation that the performance of dominant points detection depends not only on the accuracy of the measure of significance, but mainly precise determination of the region of support. This solves the fundamental problem of scale factor selection encountered in various dominant point detection algorithms. The inherent nature of scale-space filtering in the procedure is addressed and the performance of the procedure is compared to those of several other dominant point-detection algorithms, using a number of examples.>

Various types of moments have been used to recognize image patterns in a number of applications. The authors evaluate a number of moments and addresses some fundamental questions, such as image representation ability, noise sensitivity, and information redundancy. Moments considered include regular moments, Legendre moments, Zernike moments, pseudo-Zernike moments, rotational moments and complex moments. Properties of these moments are examined in detail, and the interrelationships among them are discussed. Both theoretical and experimental results are presented.>

Methods are investigated for the restoration of images degraded by both blur and noise. The objective is to develop estimation strategies to deal with images that exhibit spatially varying statistics. The restoration starts with transforming the image with nonstationary statistics into an image that exhibits stationary characteristics. This transformation can be viewed as a prewhitening filter that normalizes the local mean and local variance of the image, creating a stationary, or near stationary, field. Then the ideal image is estimated from the transformed image on the basis of the linear minimum-mean-square-error criterion. The process removes image blur and noise and at the same time inverts the effects of the transformation.>

A local feature-aggregation method for recognizing two-dimensional objects based on their CAD models is presented. The method can handle cases in which the objects are translated, rotated, scaled and occluded, and it is well suited for parallel implementation. Two types of local features, the L structures and the U structures, are extracted from the input image and matched with those of a model to search for an object similar to the model. Each of the matches hypothesizes the locations of the object in the input image, and score (similarity measure) is computed and associated with the hypothesized location to indicate the probability of the match. Matches that hypothesize the same location will have the score associated with the location incremented. A cluster of hypothesized locations with high scores indicates the probable existence of the object in the input image.>

Various types of moments have been used to recognize image patterns in a number of applications. A number of moments are evaluated and some fundamental questions are addressed, such as image-representation ability, noise sensitivity, and information redundancy. Moments considered include regular moments, Legendre moments, Zernike moments, pseudo-Zernike moments, rotational moments, and complex moments. Properties of these moments are examined in detail and the interrelationships among them are discussed. Both theoretical and experimental results are presented.>

Digital image restoration of monochrome images has been studied extensively, however, problems associated with the restoration of multichannel images still remain to be investigated. In this paper, the Wiener solution of a multichannel restoration system is presented. Using block Toeplitz to block circulant approximation, the inversion of the multichannel imaging system becomes feasible by utilizing a fast iterative matrix inversion procedure. The restoration uses both the within-channel and between-channel correlation, hence the restored result is a better estimate than the one produced by independent channel restoration. Simulations are also presented.

The constrained optimization problem for image restoration, utilizing incomplete information and partial constraints, is formulated using nonlinear programming techniques. This method restores a distorted image by optimizing a chosen object function subject to available constraints. The penalty function method of nonlinear programming is used. Both linear or nonlinear object function, and linear or nonlinear constraint functions can be incorporated in the formulation. This formulation provides a generalized approach to solve constrained optimization problems for image restoration. Experiments using this scheme have been performed. The results are compared with those obtained from other restoration methods and the comparative study is presented.

A constrained iterative image restoration method is applied to multichannel diffraction-limited imagery. This method is based on the Gerchberg-Papoulis algorithm utilizing incomplete information and partial constraints. The procedure is described using the orthogonal projection operators which project onto two prescribed subspaces iteratively. Its properties and limitations are presented. The effect of noise was investigated and a better understanding of the performance of the algorithm with noisy data has been achieved. The restoration scheme with the selection of appropriate constraints was applied to a practical problem. The 6.6, 10.7, 18, and 21 GHz satellite images obtained by the scanning multichannel microwave radiometer (SMMR), each having different spatial resolution, were restored to a common, high resolution (that of the 37 GHz channels) to demonstrate the effectiveness of the method. Both simulated data and real data were used in this study. The restored multichannel images may be utilized to retrieve rainfall distributions.

The classification of large dimensional data sets arising from the merging of remote sensing data with more traditional forms of ancillary data causes a significant computational problem. Decision tree classification is a popular approach to the problem. This type of classifier is characterized by the property that samples are subjected to a sequence of decision rules before they are assigned to a unique class. If a decision tree classifier is well designed, the result in many cases is a classification scheme which is accurate, flexible, and computationally efficient. This correspondence provides an automated technique for effective decision tree design which relies only on a priori statistics. This procedure utilizes canonical transforms and Bayes table look-up decision rules. An optimal design at each node is derived based on the associated decision table. A procedure for computing the global probability of correct classification is also provided. An example is given in which class statistics obtained from an actual Landsat scene are used as input to the program. The resulting decision tree design has an associated probability of correct classification of 0.75 compared to the theoretically optimum 0.79 probability of correct classification associated with a full dimensional Bayes classifier. Recommendations for future research are included.

This paper surveys publications, reports, and articles dealing with automated visual inspection for industry. The references are organized according to their contents: overview and discussions, rationales, components and design considerations, commercially available systems, applications. A number of applications and their inspection methodologies are discussed in detail: the inspection of printed circuit boards, photomasks, integrated circuit chips. Other inspection applications are listed as a bibliography. A list of selectively annotated references in commercially available visual inspection tools is also included.