Showing papers in "Pattern Recognition Letters in 1983"
••
TL;DR: This paper is concerned with the inexact matching of attributed, relational graphs for structural pattern recognition and the matching procedure is based on a state space search utilizing heuristic information.
474 citations
••
TL;DR: Algorithms for automatic thresholding of grey levels (without reference to histogram) are described using the terms 'index of fuzziness' and 'entropy' of a fuzzy set to be minimum when the crossover point of an S-function corresponds to boundary levels among different regions in image space.
148 citations
••
TL;DR: The proposed learning procedure consists in iterative finding such k and W which minimize the error rate estimate by the leaving 'leaving one out' method.
115 citations
••
TL;DR: A strategy is proposed for this case, by construction of convex envelopes and by the use of 'graduated' non-convexity, for adjusting a set of real-valued labels to satisfy certain constraints.
77 citations
••
TL;DR: This letter describes algorithms for global thresholding of grey-tone images which use second-order grey level statistics which are seen to be independent of the grey level histogram and effective in selecting thresholds for images with unimodal grey level distributions.
74 citations
••
TL;DR: Application to a vehicle image shows that a reconstructed image having <10% error may be obtained by using invariants formed from moments uop to order eight.
72 citations
••
TL;DR: A test of randomness based on the edge length distribution of the Minimal Spanning Tree is presented and it is shown that randomness is higher on the right side of the distribution than the left.
64 citations
••
TL;DR: Two new image-coding systems are proposed, in which a one-dimensional Hadamard transform is applied on each row of image data, and a vector-quantization process along the columns of the transformed image data.
47 citations
••
TL;DR: A method for handprinted Kanji character classificatin is proposed and a feature vector that represents the distribution of strokes is generated and is matched with average vectors in a dictionary.
44 citations
••
TL;DR: The n-tuple recognition net is seen as a building brick of a progression of network structures and the emergent 'intelligent' properties of such systems are discussed.
36 citations
••
TL;DR: Some simple properties of the planar Euclidean relative neighbourhood graph are derived from the Gabriel graph and a comparison of analogous properties is given.
••
TL;DR: This work analyzes (on pseudo-randomly generated data) the errors committed with a log(N) algorithm, attempting to recognize the nearest neighbour among N vectors, and proposes improvements in O(log(N), and presents an example of application on speech data.
••
TL;DR: An algorithm is presented which obtains a constrained maximum likelihood classification of homogeneously stained chromosomes and extension to banded chromosomes and abnormal cells are discussed.
••
TL;DR: A method of taking 3D information into account in the segmentation process is introduced, where the image intensities are adjusted to compensate for the effects of estimated surface orientation; the adjusted intensities can be regarded as reflectivity estimates.
••
TL;DR: It is shown in this paper that the minimum distance between two finite planar sets of n points can be computer in O(n log n) worst-case running time and that this is optimal to within a constant factor.
••
TL;DR: Two hierarchical classifiers strategies which minimize the global and local probabilities of misclassification, respectively are presented and the modified version of the k-NN rule for a hierarchical classifier is proposed.
••
TL;DR: This paper describes a segmentation scheme applied to Synthetic Aperture Radar images where an edge-perserving smoothing operator is used, followed by a bonding technique whereby a similarity measure is applied to each pair of neighbouring pixels.
••
TL;DR: It is shown that the first step does not always work correctly and can even yield non-simple polygons, invalidating the use of the second step of the two-stage convex hull algorithm.
••
TL;DR: A formula to calculate the length of a path between two arbitrary voxels of a digitized three-dimension (3D) picture is presented and distance functions on the 3D digitized space are derived.
••
TL;DR: Powerful univariate and multivariate normality tests are surveyed and a few transformation methods which can be used for classification purposes in case of non-normality are discussed.
••
TL;DR: If G represents an adjacency relation between points in a digital picture (or higher-dimensional array), this shows that the connected components of points can be counted by computing the rank of E, the incidence matrix of a graph G.
••
TL;DR: The shape of an open curve is defined which does not depend on its position, size and orientation and two distance measures between shapes are constructed on the basis of which two shape similarity measures are defined.
••
[...]
TL;DR: An algorithm is presented which finds the best-fitting pair of constants, in the least squares sense, to a set of scalar data, called the 'bimean' of the data.
••
TL;DR: A new class of metrics, called path-generated metrics, is defined for the digital plane, including the classical city block and chessboard metrics.
••
TL;DR: The probability of misclassification, conditioned on a learning sequence of length n, is shown to converge to the Bayes risk almost surely as rapidly as O(n^-^1^2^+^@d), @d positive.
••
TL;DR: A computer simulation for handprinted Chinese characters by extracting and composing an ordered sequence of strokes is done, and the results are satisfactory.
••
TL;DR: The Generalized Hough Transform is an established technique for geometric shape matching that can be used to find partial isomorphisms and that it can be readily implemented in parallel on a network of simple processors.
••
TL;DR: It is shown that a geometric transform has this intersection property if and only if it is defined in a special way in terms of a 'neighborhood base', which is a generalization of the class of distance transforms.
••
TL;DR: An improved stroke matching method for the recognition of handprinted Kanji characters using the stroke feature and two additional global features with a recognition rate over 90% has been obtained.
••
TL;DR: The design and construction of a new image processing system, CLIP7, is described, together with the design and operation of the custom integrated circuit on which it is based.