Showing papers on "Feature vector published in 1969"

Experiments with "Characteristic Loci" for Recognition of Handprinted Characters

Abstract: Recognition experiments have been performed on handprinted characters using a set of features which has not been applied previously to handprinting. Glucksman's "characteristic loci" were utilized in experiments with the well-known Highleyman data, as well as samples generated at Stanford Research Institute and Honeywell. Two recognition algorithms were tested. Results on numeric samples compare favorably with those of other investigators despite the small dimensionality of the feature vector. On the constrained Honeywell samples, recognition rates exceeding 98 percent were achieved using the simpler algorithm. With alphabetic samples, some problems remain in resolving persistent ambiguities, and methods for attacking these problems are considered.

Feature Extraction on Binary Patterns

Abstract: The objects and methods of automatic feature extraction on binary patterns are briefly reviewed. An intuitive interpretation for geometric features is suggested whereby such a feature is conceived of as a cluster of component vectors in pattern space. A modified version of the Isodata or K-means clustering algorithm is applied to a set of patterns originally proposed by Block, Nilsson, and Duda, and to another artificial alphabet. Results are given in terms of a figure-of-merit which measures the deviation between the original patterns and the patterns reconstructed from the automatically derived feature set.

Randomly Generated Nonlinear Transformations for Pattern Recognition

Abstract: In many mathematical and engineering problems the solution is simpler after a transformation has been applied. A general method is proposed to find suitable transformations for discrete data in information processing problems. The main feature of the method is random perturbation of the data subject to constraints which ensure that, in the transformed space, the problem is in some sense simpler and that the local structure of the data is preserved. An application of this technique to pattern recognition is discussed where a transformation is found for the feature space such that classes, which are not linearly separable in the original space, become so in the transformed space. The transformation considerably simplifies the problem and allows well-developed linear discriminant techniques to be applied. This application was implemented and tested with a number of examples which are described.

Nonorthogonal projections for feature extraction in pattern recognition

Abstract: It is known that R linearly separable classes of multi-dimensional pattern vectors can always be represented in a feature space of at most R dimensions. An approach is developed which can frequently be used to find a non-orthogonal transformation to project the patterns into a feature space of considerably lower dimensionality. Examples involving classification of handwritten and printed digits are used to illustrate the technique.