Showing papers in "Pattern Recognition Letters in 1993"

TL;DR: It is shown that setting F= 1 is the more appropriate for ellipse fitting since it is less heavily curvature biased and less well suited than the A + C= 1 normalisation with respect to singularities and transformational invariance.

TL;DR: This work uses a genetic algorithm to find a near-optimal partitioning of the given data set by selecting proper initial seed values in the K-means algorithm, and results obtained are very encouraging.

TL;DR: This work suggests variants for 2, 3 and arbitrary dimensions that are separable, suitable for various parallel architectures, and presents a 4-scan algorithm for 3-dimensional images.

TL;DR: Several recognition algorithms used in the interpretation of handwritten and machine-printed address text (digits/symbols/alphabets/words) are described.

TL;DR: A method that combines clustering and region merging for color image segmentation using recursive one-dimensional histogram analysis and a criterion that takes into account color similarity and spatial proximity is presented.

TL;DR: An assessment of the current state of the art in handwriting recognition is given and some evidences and novel ideas on ways of stretching the limits of handwriting recognition systems aiming at outperforming human beings are presented.

TL;DR: A simple but efficient method is proposed for detection of significant points of chain-coded curves by joining successive significant vertices through the polygonal approximation.

TL;DR: Standard techniques for ellipse fitting such as the least squares method and the Hough Transform either make certain assumptions about the type of noise distribution, or require input parameters to work robustly over a large range of data, are described.

TL;DR: A necessary and sufficient condition for a fuzzy graph to be isomorphic to its corresponding fuzzy line graph of some fuzzy graph is given.

TL;DR: An algorithm that combines region growing and edge detection to segment out an object in the image and uses gradient information to detect the optimal region boundary and stop the growth.

TL;DR: This letter describes an approach to characterize texture orientation based on directional evidence accumulation as in the Hough Transform, and a measure to grade the directional image is defined.

TL;DR: A new fuzzy clustering algorithm called the Fuzzy C Quadric Shells algorithm is introduced which is expressly designed to seek clusters that can be described by segments of second-degree curves, or more generally by segment of shells of hyperquadrics.

TL;DR: In this article, the Hough transform is applied over the source image and then the slope of handwritten rows is estimated by analyzing the parameter plane, which is used for cutting the image and forming "strips" containing the respective handwritten rows.

TL;DR: An equivalence is proved between the degree of connectedness defined for fuzzy sets and the connection cost defined in the grey-level mathematical morphology framework, starting from the definitions and properties of both concepts.

TL;DR: It is shown that an ear of P can be found in linear time with a simple, practically efficient algorithm that does not require pre-triangulating P.

TL;DR: This paper discusses measures for polygons that may be computed efficiently, and in which closeness implies similarity with respect to deviation from convexity.

TL;DR: An idea is presented for on-line change of the recognition rule in dependence on the reliability of the classification decision, aiming at better accuracy.

TL;DR: It is shown that the modified weighted sum of square error function can be optimized to obtain expressions for the parameters of the circular fit so that expensive iterative numerical optimization algorithms are not needed.

TL;DR: Current efforts to deal with OCR as a signal processing problem where the causes of noise and distortions as well the idealized images (definitions of typefaces) are modeled and subjected to a quantitative analysis.

TL;DR: A hashing-oriented nearest neighbor searching scheme given n points in the Euclidean two-dimensional plane is presented, which compares the distance between the query point and the dominated points to determine the nearest neighbor.

TL;DR: A recent segmentation method that uses transition regions to select an appropriate threshold for binarization is compared to an older averaged gradient thresholding method and it is shown that these methods are closely related to each other.

TL;DR: Two methods to compute the EN of a binary image by means of the Connectivity Graph, which is derived from the Cylindrical Algebraic Decomposition of the Euclidean plane, are shown.

TL;DR: Experimental results show that the new edge detector performs better than standard morphological methods, especially in the presence of noise.

TL;DR: A survey of the research projects conducted at the French post office research center (SRTP) on the recognition of printed and handwritten mailing addresses for small envelopes and flat mail and on the Recognition of postal check values is given.

TL;DR: A dynamic alignment algorithm is used to compute the similarity index between two sets of shape descriptors, which is hierarchical and is invariant to rotation, translation and scaling.

TL;DR: A novel methodology for shape discrimination is introduced by combining pattern recognition techniques such as morphological processing with concepts from artificial intelligence and machine learning such as genetic algorithms (GAs) to develop robust shape analysis methods.

TL;DR: The dimensionality of feature vectors is reduced by using the principles of Karhunen-Loeve transform applied to the feature images locally and globally by choosing the resulting basis vectors which are closest to the classical KL basis vectors.

TL;DR: This paper proves that digital line segments and their least square line fits are in one-to-one correspondence and gives a new simple representation of a digital line segment, where x 1 and n are the x -coordinate of the left endpoint and the number of digital points.

TL;DR: A new method for cursive script recognition based on the extraction of key letters which consist of the parts of the handwritten text which are singularities of handwriting, and the problems leading to the missegmentation are identified for further improvement.

TL;DR: Two methods for the detection of ellipses based on the straight line Hough transform (SLHT) are presented and timing performance of the proposed methods compares favorably with the other Hough-based methods.