Showing papers in "Pattern Recognition in 1979"
••
TL;DR: This paper provides a semi-tutorial review of the state-of-the-art in cluster validity, or the verification of results from clustering algorithms, and covers ways of measuring clustering tendency, the fit of hierarchical and partitional structures and indices of compactness and isolation for individual clusters.
298 citations
••
TL;DR: A polarogram is a polar plot of an orientation sensitive texture statistic which gives rise to a class of texture descriptors which are sensitive to both texture coarseness and directionality, but yet which are invariant to rotations of the image textures.
110 citations
••
TL;DR: A method for describing scenes with polyhedra and curved objects from three-dimensional data obtained by a range finder that might be useful for the recognition of the objects.
90 citations
••
TL;DR: Results obtained indicate this two-dimensional image model is formulated using a seasonal autoregressive time series and could be used to code textures for low bit rates or be used in an application of generating compressed background scenes.
87 citations
••
TL;DR: Some attempts to segment textured black and white images by detecting clusters of local feature values and partitioning the feature space so as to separate these clusters.
75 citations
••
TL;DR: A new approach to problems of clustering and classification of multidimensional pictorial data is presented and the development of a clustering technique and program is described.
71 citations
••
TL;DR: The formal definitions of both the definite and the definite hierarchical clustering procedure with the dissimilarity coefficient D are given, by means of which the properties of these procedures can be investigated.
71 citations
••
TL;DR: Methods of fitting polygons and arcs of ellipses to the border sequences are developed and applied to finding small holes in images of connecting rods.
70 citations
••
TL;DR: It is shown that Bayesian probability measures, which are used for feature selection, and are based on distance measures and information measures, are basically of two types, which clarifies some properties of these measures for the two-class problem and for the multiclass problem.
60 citations
••
TL;DR: The methods used to describe the boundaries of the figures, to match common shapes between paired images, and then to analyze the motion of matched figures, are discussed in detail.
56 citations
••
TL;DR: The quadtree representation encodes a 2″ by 2″ binary image as a set of maximal blocks of 1's or 0's whose sizes and positions are powers of 2, and a hierarchy of approximations to the image can be defined.
••
TL;DR: Comparisons with principal component analysis, hierarchical classification, and other methods, show why the P.A.P. method (Progressive Assimilation and Percolation) is superior and safer than most known methods.
••
TL;DR: A statistical analysis of the variations of this system on a set of successive pages leads to the extraction of writer features as well as the segmentation of the text into short-time stationarity domains to be related to “rhythms of writing”.
••
TL;DR: This paper selectively surveys contributions in linear feature selection which have been developed for the analysis of multipass LANDSAT data in conjunction with the Large Area Crop Inventory Experiment.
••
TL;DR: The theory of Latin Square experimental designs is extended to edge detection of multi-grey level pictorial data and post hoc comparison method is used to confine the edge element ambiguities to 2-pixel layer thickness in masks greater than 2 × 2 × k.
••
TL;DR: A new data compression algorithm for a binary image which has a hierarchical format for describing a macroscopic image structure and can be applied for the image data storage, image edition and image transmission.
••
TL;DR: The effect of learning sample size on the optimal pattern recognition dimensionality is considered and some procedures for determination of the optimal dimensionality are described and compared by a simulation method.
••
TL;DR: The similarity between the single-link clustering method and a recently proposed clustering methods based on mode-seeking is demonstrated.
••
TL;DR: Several software engineering techniques are proposed for the implementation of a more efficient Error-Correcting Tree Automata (ECTA), and a parallel parsing algorithm of tree languages is introduced.
••
TL;DR: A simple, intuitive, nonparametric clustering procedure, based on such overlapping pattern-cells is presented and may be classified as an agglomerative, hierarchical, linkage-type clustering process.
••
TL;DR: Tests of specific “recognition cone” systems for probabilistic parallel-serial recognition and description of two-dimensional scenes of objects showed a wide variety of diverse sources of information contextually interact, in a relatively simple and general way.
••
TL;DR: The paper illustrates the variety of texture which can be produced by the method, which arranges the generated gray tones in a sequence along a scan line, using a regular Markov chain.
••
TL;DR: Techniques for calculating the stroke directions of thinned binary characters and for detecting the intersections and end points of strokes by means of pattern matching and weighting method are proposed as a preprocessing of handwritten Chinese character recognition.
••
TL;DR: Test results from a series of holdout experiments indicate that average correct recognition rates of about 85% can be achieved on the atypical cells, while maintaining an error rate of about 1% on the normal cells.
••
TL;DR: A model to reduce the dimension of the finite dimensional data is developed for the case when the covariance matrices are not necessarily equal and sufficient conditions are given so that the linear transformation of data of higher dimensions to lower dimensions does not increase the probabilities of misclassification.
••
TL;DR: A new method for linear feature selection which has as its underlying theme the preservation of actual distances between training data points in the lower dimensional space and places the method closer to the principle components or Karhunen- Loeve approach than to methods based on an approach through statistical pattern recognition.
••
TL;DR: The concept of a “mutualistic teacher” is introduced for unsupervised learning of the mean vectors of the components of a mixture of multivariate normal densities, when the number of classes is also unknown.
••
TL;DR: Experiments in the detection of parallel sided strips using a relaxation-like process which iteratively reinforces collinear or anti-parallel edges using two types of data, tree trunks and runways.
••
TL;DR: A group structure is identified among the Regular Kolam Array Grammars that generate the families of Kirsch's right triangles with labelled vertices and their reflections about the leg and/or base of the triangles.
••
TL;DR: This paper refutes the assertion of Shaffer et al. that, in general, the minimum spanning tree algorithm and the mode-seeking clustering algorithm yield identical results.