Theodosios Pavlidis

Bio: Theodosios Pavlidis is an academic researcher from State University of New York System. The author has contributed to research in topics: Segmentation & Image segmentation. The author has an hindex of 41, co-authored 93 publications receiving 10089 citations. Previous affiliations of Theodosios Pavlidis include Princeton University & Symbol Technologies.

Syntactic Analysis of Region Boundaries and Other Curves

Abstract: Encodings of curves as sequences of arcs are particularly suited for syntactic descriptions. If each arc is labeled by a symbol from a finite alphabet, then the curve is mapped into a string over such an alphabet and the classical theory of formal languages [8.1] is directly applicable. Much of the early work in syntactic pattern recognition has used representations based on the chain code or its variants (for a review see [8.2]). A recent example of its use can be found in the work of JARVIS [8.3]. Under this formalism each symbol corresponds to an arc of very short length and the symbol name signifies the direction of a linear segment approximating the arc (see Sec.7.5). This has the advantage of obtaining encodings very easily, but it places the burden for noise removal on the syntactic analyzer. It also requires the use of syntactic methodology for checking relatively simple geometrical properties, like length equality. These disadvantages may be overcome if we preprocess the input curves and obtain encodings in terms of symbols which represent higher order structures rather than small arc segments. Piecewise polynomial approximations offer one possibility, since they remove much of the noise and make information regarding simple properties, like length, readily available. However, they cannot be represented in terms of a finite alphabet, and, for certain applications, they may offer an encoding which is still too “elementary”. In this chapter we shall describe techniques which overcome these problems and produce curve descriptions in terms of finite alphabets, but with each symbol representing a more complex part of a curve than a short arc or an arc which is nearly linear.

Classification, Description and Syntactic Analysis

Abstract: By this time we have seen how a picture (or a waveform) can be transformed into a vector (e.g., one whose components are the Fourier descriptors), a string of symbols (e.g., the chain code), or a graph (e.g., the RAG). We are faced now with the problems of description and/or classification. Description requires that a text in a natural language, or a language which can be translated easily into such, be produced which contains all relevant information about a picture. Classification requires the assignment of a picture (or parts of it) to one of a finite number of classes. In many cases the two processes overlap. For example, in order to describe a picture of a room we may have to classify first the various objects in it. Conversely, a number of classification techniques are based on descriptions, as it was shown to be the case with the examples of Section 8.6. One may even go as far as to state that structural pattern recognition is a sequence of alternating classifications and descriptions: A pixel is classified as belonging to one of several regions. These regions provide a rudimentary description of the picture. Parts of each region are classified as belonging to its boundary or to some of its simpler components. Then a region may be described through its parts, or its boundary may be described as a sequence of arcs, etc.

Learning algorithms when class membership is poorly defined

Abstract: Let X be a measurement space and f(x) a real function defined on it If f(x) takes only a small set of discrete values then we have the standard classification problem Otherwise f (x) can be considered as defining a fuzzy pattern recognition problem We consider the problem of dividing X into regions xi( = 1,2 R) such that on each one of them f(x) is approximated either by a constant or a linear function The partition is generated for a given R by minimizing the total integral square error This problem is equivalent to piecewise functional approximation After the regions Xi and the approximations have been determined than it is possible to predict the value f(x) for any given measurement x The computational requirements of this approach are higher than those of the common learning algorithms but it is applicable in cases where (except for extreme cases) class membership is vaguely defined as it is often the case in socio-economic problems, mechanical and medical diagnosis etc

Statistical pattern recognition: a review

Abstract: The primary goal of pattern recognition is supervised or unsupervised classification. Among the various frameworks in which pattern recognition has been traditionally formulated, the statistical approach has been most intensively studied and used in practice. More recently, neural network techniques and methods imported from statistical learning theory have been receiving increasing attention. The design of a recognition system requires careful attention to the following issues: definition of pattern classes, sensing environment, pattern representation, feature extraction and selection, cluster analysis, classifier design and learning, selection of training and test samples, and performance evaluation. In spite of almost 50 years of research and development in this field, the general problem of recognizing complex patterns with arbitrary orientation, location, and scale remains unsolved. New and emerging applications, such as data mining, web searching, retrieval of multimedia data, face recognition, and cursive handwriting recognition, require robust and efficient pattern recognition techniques. The objective of this review paper is to summarize and compare some of the well-known methods used in various stages of a pattern recognition system and identify research topics and applications which are at the forefront of this exciting and challenging field.

Efficient Graph-Based Image Segmentation

Abstract: This paper addresses the problem of segmenting an image into regions. We define a predicate for measuring the evidence for a boundary between two regions using a graph-based representation of the image. We then develop an efficient segmentation algorithm based on this predicate, and show that although this algorithm makes greedy decisions it produces segmentations that satisfy global properties. We apply the algorithm to image segmentation using two different kinds of local neighborhoods in constructing the graph, and illustrate the results with both real and synthetic images. The algorithm runs in time nearly linear in the number of graph edges and is also fast in practice. An important characteristic of the method is its ability to preserve detail in low-variability image regions while ignoring detail in high-variability regions.

The Graph Neural Network Model

Abstract: Many underlying relationships among data in several areas of science and engineering, e.g., computer vision, molecular chemistry, molecular biology, pattern recognition, and data mining, can be represented in terms of graphs. In this paper, we propose a new neural network model, called graph neural network (GNN) model, that extends existing neural network methods for processing the data represented in graph domains. This GNN model, which can directly process most of the practically useful types of graphs, e.g., acyclic, cyclic, directed, and undirected, implements a function tau(G,n) isin IRm that maps a graph G and one of its nodes n into an m-dimensional Euclidean space. A supervised learning algorithm is derived to estimate the parameters of the proposed GNN model. The computational cost of the proposed algorithm is also considered. Some experimental results are shown to validate the proposed learning algorithm, and to demonstrate its generalization capabilities.

On combining classifiers

Abstract: We develop a common theoretical framework for combining classifiers which use distinct pattern representations and show that many existing schemes can be considered as special cases of compound classification where all the pattern representations are used jointly to make a decision. An experimental comparison of various classifier combination schemes demonstrates that the combination rule developed under the most restrictive assumptions-the sum rule-outperforms other classifier combinations schemes. A sensitivity analysis of the various schemes to estimation errors is carried out to show that this finding can be justified theoretically.

A survey of image registration techniques

Abstract: Registration is a fundamental task in image processing used to match two or more pictures taken, for example, at different times, from different sensors, or from different viewpoints. Virtually all large systems which evaluate images require the registration of images, or a closely related operation, as an intermediate step. Specific examples of systems where image registration is a significant component include matching a target with a real-time image of a scene for target recognition, monitoring global land usage using satellite images, matching stereo images to recover shape for autonomous navigation, and aligning images from different medical modalities for diagnosis.Over the years, a broad range of techniques has been developed for various types of data and problems. These techniques have been independently studied for several different applications, resulting in a large body of research. This paper organizes this material by establishing the relationship between the variations in the images and the type of registration techniques which can most appropriately be applied. Three major types of variations are distinguished. The first type are the variations due to the differences in acquisition which cause the images to be misaligned. To register images, a spatial transformation is found which will remove these variations. The class of transformations which must be searched to find the optimal transformation is determined by knowledge about the variations of this type. The transformation class in turn influences the general technique that should be taken. The second type of variations are those which are also due to differences in acquisition, but cannot be modeled easily such as lighting and atmospheric conditions. This type usually effects intensity values, but they may also be spatial, such as perspective distortions. The third type of variations are differences in the images that are of interest such as object movements, growths, or other scene changes. Variations of the second and third type are not directly removed by registration, but they make registration more difficult since an exact match is no longer possible. In particular, it is critical that variations of the third type are not removed. Knowledge about the characteristics of each type of variation effect the choice of feature space, similarity measure, search space, and search strategy which will make up the final technique. All registration techniques can be viewed as different combinations of these choices. This framework is useful for understanding the merits and relationships between the wide variety of existing techniques and for assisting in the selection of the most suitable technique for a specific problem.