Showing papers in "Pattern Recognition in 1991"

TL;DR: A texture segmentation algorithm inspired by the multi-channel filtering theory for visual information processing in the early stages of human visual system is presented, which is based on reconstruction of the input image from the filtered images.

TL;DR: It is shown that if just a small subset of the edge points in the image, selected at random, is used as input for the Hough Transform, the performance is often only slightly impaired, thus the execution time can be considerably shortened.

TL;DR: The authors show that the new method of deriving Zernike moment invariants along with the new normalization scheme yield the best overall performance even when the data are degraded by additive noise.

TL;DR: The simulated annealing approach for solving optimization problems is described and is proposed for solving the clustering problem and the parameters of the algorithm are discussed in detail and it is shown that the algorithm converges to a global solution of the clustered problem.

TL;DR: A new dissimilarity measure, based on “position”, “span” and “content” of symbolic objects is proposed for symbolic clustering, and the results of the application of the algorithm on numeric data of known number of classes are described first to show the efficacy of the method.

TL;DR: This paper proves that SV feature vector has some important properties of algebraic and geometric invariance, and insensitiveness to noise, and these properties are very useful for the description and recognition of images.

TL;DR: The authors combine two algorithms for application to the recognition of unconstrained isolated handwritten numerals utilizing features derived from the profile of the character in a structural configuration to recognize the numerals.

TL;DR: It is proved that the generalized eigenequations on the optimal discriminant plane are stable in respect of eigenvalues and the generalized Eigenvectors are indeed the optimal discriminate directions, if the perturbation is subject to some certain conditions.

TL;DR: A new set of textural measures derived from the texture spectrum is presented, which extract textural information of an image with a more complete respect of texture characteristics (in all the eight directions instead of only one displacement vector used in the co-occurrence matrix approach).

TL;DR: A system which can recognize digits and uppercase letters handprinted on a touch terminal is described, analogous to “time delay neural networks” previously applied to speech recognition.

TL;DR: Experimental results show that dominant points obtained by this method are less sensitive to the orientation of the boundary than other polygonal approximation algorithms in the sense that the number and the location of the dominant points along the contour remain relatively unchanged.

TL;DR: An iterative algorithm is proposed for moment calculation which needs no multiplications, and the number of additions needed is reduced to O ( N ), which shows that the computational complexity is significantly reduced.

TL;DR: A new method of recognizing handwritten Chinese characters using a structural representation called hierarchical attributed graph representation (HAGR) which can tolerate the variations of HAGR which reflect the instabilities or variabilities of handwrittenChinese characters resulting from different writing styles.

TL;DR: A method of retrieving iconic pictures in a pictorial database based upon the spatial relationship among the objects in the picture is suggested and the pictures satisfying that query can be easily determined.

TL;DR: An approach in which the statistical and structural approaches are combined to represent the fingerprint false minutia patterns is presented, and a fingerprint image postprocessing algorithm is developed to eliminate thefalse minutiae in fingerprint images.

TL;DR: A simple iterative algorithm is proposed for the computation of moments from a polygonal approximation of the boundary that can be applied to compute the most popular geometric moments as well as other types of moments like Legendre, Zernike, rotational and complex moments.

TL;DR: The feasibility of using isodensity lines for human face identification system is presented through experimental investigation and shows a 100% accuracy in matching same persons and a 100%" accuracy in discriminating different persons (including persons with spectacles).

TL;DR: The time complexity of the shape matching algorithm is O(m), where m is the number of feature points, and the most important advantage of the method is its independence to translation, rotation and scaling of objects.

TL;DR: An efficient algorithm for agglomerative clustering using a heap in which distances of all pairs of clusters are stored and updating the heap at each stage of the hierarchy is easily implemented.

TL;DR: A method to represent polygons as cyclic strings is proposed and it is shown how cyclic string-matching techniques can be used for rotation-, translation- and scale-independent polygonal shape recognition.

TL;DR: A method which computes the same set of moments using only the corner pixels along a shape's boundary using the basic approach is to construct a set of triangles using the shape's corners and the origin of the coordinate system.

TL;DR: Unlike other dominant point detection algorithms which are sensitive to scaling and rotation of the object curve, the new method will overcome this difficulty and is robust in the presence of noise.

TL;DR: A pipeline strategy for handwritten numeral recognition that combines a two-stage template-based technique and a model- based technique is described, which classifies 70–80% of the digits with reliability rates over 99% and generates class membership hypotheses for the remaining digits which constrain the model-based system.

TL;DR: A gesture input filter and recognizer is built, a novel combination of two traditional techniques; angle filtering and multiscale recognition.

TL;DR: Automatic detection of the principal axes of given shapes with known symmetry properties is studied and analytic equations for computing the direction of high-order principal axes are derived.

TL;DR: The implementation method that is used is significantly superior to the classical ones in some aspects (e.g. computational time, memory storage) and experimental results show that the use of IIR filter preserves the detector properties.

TL;DR: The classifier is designed on the base of the learning sets in such a manner that these sets are completely reclassified by a two-layer network of formal neurons.

TL;DR: Techniques for decomposing big grayscale morphological structuring elements into combined structures of segmented small components will allow full freedom for users to design any kind and any size of gray-scale morphologicalStructuring element.

TL;DR: Two different interpolative schemes for adaptive contrast enhancement have been presented for speeding up the contrast enhancement algorithm proposed by Beghdadi and Negrate and a contrast transformation function is proposed which can be used for adaptive enhancement as well as de-enhancement of images.

TL;DR: This paper studies the properties of a set of four related closest-point graphs using Monte Carlo methods and finds Delaunay triangulation to be the least sensitive to such noisy conditions.