Showing papers in "Pattern Recognition Letters in 1986"

TL;DR: A new algorithm is proposed which finds the Nearest Neighbour of a given sample in approximately constant average time complexity, independent of the data set size, thus being of general use in many present applications of Pattern Recognition.

TL;DR: It is shown that a ‘continuous’ function from a finite block of lattice points into itself has an ‘almost-fixed’ point P such that f ( P ) is a neighbor of P (diagonal neighbors allowed).

TL;DR: This report describes a method of detecting the orientation of a set components which are located along parallel lines by using the histogram of nearest neighbor directions.

TL;DR: Different types of error rates are described and the state of the art of error rate estimation at the time of Toussaint's survey is briefly summarised, and the two major advances, namely bootstrap and average conditional error rates estimation methods, and their extensions, are described in detail.

TL;DR: A new approach to the problem of detecting and enhancing linear features in noisy digital images is described, based on the Radon transform, which can be efficiently implemented on a general-purpose digital computer.

TL;DR: An algorithm based on the facts of the human visual system is presented here whereby it is possible to select automatically (without human intervention) the thresholds for detecting the significant edges as perceived by human beings.

TL;DR: This paper reviews a few algorithms from the literature which formed the basis for a procedure with the following properties: it generates a binary decision tree, it can handle mixed data types, it performs the feature and criterion selection simultaneously, and it is interactive and hence enables the user to control the design procedure.

TL;DR: A computationally efficient technique, based on the Hough transform, for the detection of straight edges in industrial images, which permits objects to be located within 1 pixel and orientated within ∼1°.

TL;DR: Two membership functions in R have been defined such that the fuzzy measures posses maximum values when the crossover points/central points of the membership functions correspond to the points around which the data has a tendency to cluster.

TL;DR: The 4-metric is adopted to construct the Voronoi diagram of a binary digital picture, whose foreground consists of arbitrarily shaped components and the various tiles are identified by using a component labeling technique.

TL;DR: The concept of circularity is shown to be important for the design of 'general purpose' template matching operators, and the underlying theory with particular reference to computational accuracy is developed.

TL;DR: Automatic methods of extracting information on large numbers of stars and galaxies are described with emphasis on the image processing techniques employed.

TL;DR: A scalar transform to describe a 2-dimensional shape does not depend on the position, size and orientation of a region, and it is shown how the shape vector can be used in discriminating shapes.

TL;DR: A stereo matching algorithm based on dynamic programming that produces a set of candidate matches for each row separately and uses continuity constraints to choose the best matches.

TL;DR: The algorithm combines results from visibility and shortest paths with the linear-time polygon triangulation algorithm discovered recently by Tarjan and Van Wyk to solve the strong hidden-line problem in a simple polygon P.

TL;DR: It is shown that with O(n log n ) preprocessing of P, an edge-to-edge visibility query can be answered in O( n ) time and the algorithm reports a line-of-sight if the answer is in the affirmative.

TL;DR: It is stated, without proof, that the total fraction of samples discarded by the Multiedit algorithm is bounded from above by 2E1, where E1 is the 1-NNR error rate for the initial distributions.

TL;DR: The inherent fuzzifying property of the S-function is found to enable one not to use the INT operator for contrast enhancement and to save the time of computation greatly as measured by the index value.

TL;DR: This letter describes and compares a range of bi-level architectures, identifying some principal design factors of significance and illustrating the flexibility of such configurations in terms of operational characteristics.

TL;DR: If a chromosome classifier is constrained to produce fixed class sizes specified a priori then classification accuracy may be improved compared with a simple maximum likelihood approach applied to each chromosome without reference to the other chromosomes in the cell.

TL;DR: An image is represented as a four-layer structure consisting respectively of pixels, blocks, segments and regions that leads to recognition of anatomical structures within CT brain scan images.

TL;DR: Four algorithms for percentile filtering are compared and their flexibility, speed, and the ways they can be implemented in hardware and software are considered.

TL;DR: A novel method of image and scene segmentation is presented which detects boundaries in the image between optical flow fields from different moving planar facets in the scene.

TL;DR: A rule based boundary tracing algorithm which uses stronger prior knowledge expressed in rule form to give a richer mapping which more closely describes the structural distinctions in the scene.

TL;DR: A method is presented for the detection of a two-dimensional extended pattern of points in a flat background based on the statistical properties of a single-link clustering algorithm, which allows to select the most significant clusters.

TL;DR: Syntactic pattern recognition techniques are used to classify a small number of normal and abnormal bifurcations in carotid angiograms, which seems successful but necessitates the development of a sophisticated preprocessor if it is to be of practical value.

TL;DR: The application of linear decision functions to the determination of the optimal recognition algorithms is presented and the results are obtained in a numerical way using a random method of optimization.

TL;DR: An integrated set of data structures and image processing primitives which enhance not only the run time efficiency of the application program, but also the programmability of the solution are described.

TL;DR: Through consideration of the constraints governing the correspondence problem, a new relaxation labelling scheme is proposed and numerical examples are presented which illustrate the improved performance of the scheme.

TL;DR: Cluster analysis is shown to be capable of taking into account structural information at the cell level, which may be lost in other methods.