Showing papers in "Pattern Recognition in 1992"

TL;DR: The capability of the human visual system with respect to face identification, analysis of facial expressions, and classification based on physical features of the face are discussed.

TL;DR: It is demonstrated that integrating the information extracted from multiresolution SAR models gives much better performance than single resolution methods in both texture classification and texture segmentation.

TL;DR: Comparisons of textural features for pattern recognition show that co-occurrence features perform best followed by the fractal features, however, there is no universally best subset of features.

TL;DR: In this paper, the performance of a Gabor odd filter-based edge detector is investigated using the measures proposed by Canny and it is shown that this design criterion also holds good for a two-dimensional (2D) Gabor filter- based edge detector.

TL;DR: A new method is proposed to estimate fractal dimension in a two-dimensional (2D) image which can readily be extended to a 3D image as well.

TL;DR: An efficient scheme for two-dimensional data encryption is presented based on the principles and ideas reflected by the specification and development of the SCAN language, and is mainly motivated for the encryption of 2D digital pictures.

TL;DR: It is shown that Kittler and Illingworth's thresholding, which minimizes a criterion related to the average classification error rate assuming normal distribution with different variances, is equivalent to the maximization of the likelihood of the joint distribution in the population mixture model.

TL;DR: In an effort to use more of the information available in the image, the present approach evaluates two-dimensional entropies based on the 2D (grey-level/local average grey-level) “histogram” or scatterplot.

TL;DR: Fifty-two invariant functions of the second- to the ninth-order image moments are tabulated, which are invariant with respect to image translation, orientation and size changes and can be used as library functions in computer programs for automatic image identification.

TL;DR: The spatial query and spatial reasoning based on a 2D C-string representation are presented and the similarity measure is defined and the algorithm for similarity retrieval of iconic images is also proposed.

TL;DR: Techniques aimed at improving the efficiency and reducing the memory size of the accumulator array are presented, which show that only a 2-dimensional array is needed for the detection of circles and ellipses.

TL;DR: A compatible cluster merging algorithm is presented that is specially designed to find the optimum number of linear, planar and hyperplanar clusters when an upper bound on the number of clusters present is known.

TL;DR: Because the light pattern is encoded, the correspondence between the projected beams from the pattern and imaged beams is easily determined, and hence range data may be solved for quickly, and therefore this method may be used in dynamic environments.

TL;DR: This method makes use of structural information culled from the directional histograms of the directional image of a fingerprint to detect singular points in a fingerprint image.

TL;DR: Experiments are shown to illustrate the matching algorithm and to demonstrate that the approach is truly invariant to arbitrary translations, rotations, and scale changes under noise.

TL;DR: A detailed analysis of this pixel classification methodology using a Gaussian maximum likelihood classifier is provided, and included is a statistical study of classification accuracy, feature optimization, and robustness with respect to various relevant noise models.

TL;DR: The experimental results show that the present method is superior to the Foley-Sammon method, the positive pseudoinverse method, and the matrix rank decomposition method in terms of correct classification rate.

TL;DR: The AFCS algorithms consider hyper-ellipsoidal-shells as prototypes, hence the ability to characterize elliptical boundaries, and the generalization is achieved by allowing the distances to be measured through a norm inducing matrix that is symmetric, positive definite.

TL;DR: A new approach to the determination of a mapping function from given coordinates of control points is described, which enables the registration of images with non-linear local geometric distortions with optional accuracy.

TL;DR: A systematic method for three-dimensional (3D) model matching in robot vision by using subgraph matching techniques, which identifies subgraph isomorphisms of the object's model graph.

TL;DR: A comparison of decision trees with backpropagation neural networks for three distinct multi-modal problems: two from emitter classification and one from digit recognition, which shows that both methods produce comparable error rates but that direct application of either method will not necessarily produce the lowest error rate.

TL;DR: A simple method is presented for automatically identifying regions in envelope images which are candidates for being the destination address and the success of the texture-based segmentation algorithm for identifying address blocks is demonstrated.

TL;DR: A new hybrid method is presented that combines the scale space filter (SSF) and Markov random field (MRF) for color image segmentation and it is shown that the former is histogram-based segmentation, whereas the latter is neighborhood- based segmentation.

TL;DR: In this paper, the conception of the rank decomposition of matrices is first introduced, and a new method for calculating the Fisher optimal discriminant vector is presented which is particularly well suited to the case of a small number of samples in the sense that the scatter matrices are rank-deficient.

TL;DR: Use of the image auto-correlation property is identified for further speeding up of the proposed methods in the clustering of image data and use of the Lower triangular matrix approach to speed up Dynamic clustering is proposed.

TL;DR: Evaluation of the global approximate error provides efficient curve approximations, irrespective of shape complexity, number of sampled points, and irregularity of sampling interval.

TL;DR: An effective method of scale-space filtering with a Gaussian kernel is introduced to detect dominant points on digital curves with no input parameters, and the resultant dominant points do not change under translation, rotation and scaling.

TL;DR: A new algorithm based on the double-integral formulation is presented that can be used to calculate moments from either the run-length codes or the chain codes of shape.

TL;DR: A technique is presented that identifies the significant or natural scales in a curve that greatly simplifies any reasoning processes applied to the curve.

TL;DR: A system to recognize handwritten Chinese characters is presented, where a new efficient algorithm is proposed, based on accumulated chain codes, for line approximation, in the first stage.