scispace - formally typeset
Search or ask a question

Showing papers on "Feature extraction published in 1968"


Journal ArticleDOI
George Nagy1
01 Jan 1968
TL;DR: This paper reviews statistical, adaptive, and heuristic techniques used in laboratory investigations of pattern recognition problems and includes correlation methods, discriminant analysis, maximum likelihood decisions minimax techniques, perceptron-like algorithms, feature extraction, preprocessing, clustering and nonsupervised learning.
Abstract: This paper reviews statistical, adaptive, and heuristic techniques used in laboratory investigations of pattern recognition problems. The discussion includes correlation methods, discriminant analysis, maximum likelihood decisions minimax techniques, perceptron-like algorithms, feature extraction, preprocessing, clustering and nonsupervised learning. Two-dimensional distributions are used to illustrate the properties of the various procedures. Several experimental projects, representative of prospective applications, are also described.

317 citations


Patent
08 Nov 1968
TL;DR: In this paper, a character recognition system comprising means for scanning a character on a document field, the field being composed of plural features each of which lies in a respectively defined area on the character.
Abstract: A character recognition system comprising means for scanning a character on a document field, the field being composed of plural features each of which lies in a respectively defined area on the character. A pair of video shift registers, one for positive video and one for negative video, are provided to shift the signals therethrough. Each video shift register includes a plurality of stages for serially storing and shifting a binary quantization of a character pattern sample within said field. Feature extraction logic circuitry is provided and includes a plurality of feature detecting means to provide signals indicative of the recognition of predetermined features as the character pattern is shifted through selected stages of the video shift register. The detecting means include plural mask matrices coupled to said video shift registers. The output of the matrices are provided to encoding means for encoding the signals from the matrices into plural, multi-bit binary codes, one of such codes for each of said geographic areas. Each code serves to indicate a feature detected within the geographic area. Plural shift registers are provided coupled to the encoding means to store said codes. Decoding means are coupled to the code storing shift registers to decode the signals therefrom and to provide signals indicating the recognition of a character having the features detected. Further encoding means are provided to encode the signal indicating the recognition of a character into a binary code for use by a central processor. In order to minimize the number of components to create the masks some high usage masks are gated together and provided as inputs to other masks and the first encoding means. In addition, delay means are provided associated with each mask to duplicate sub-features without necessitating the use of separate components of each sub-feature.

38 citations


Journal ArticleDOI
TL;DR: This paper discusses the selection of mathematical features on the basis of entropy minimization and introduces the concept of extracting statistical features by the method of kernel approximation.

23 citations


Journal ArticleDOI
TL;DR: Nearest-neighbor classification is used to explain the high error rates obtained by general statistical procedures, and the minimum human error rate is estimated, and suggested as a performance standard.
Abstract: —The results of three experiments with Highleyman's hand-printed characters are reported. Nearest-neighbor classification is used to explain the high error rates (42 to 60 percent) obtained by general statistical procedures. An error rate of 32 percent is obtained by preceding piecewise-linear classification by edge-detecting preprocessing. The minimum human error rate is estimated, and suggested as a performance standard.

21 citations


Journal ArticleDOI
TL;DR: It is shown that certain important normalizations (position, size, pitch, etc.) are nonlinear operations.
Abstract: Pattern recognition (including sound recognition) is described mathematically as the problem to compute for any element of a given class its image in a classification set. The difficulty lies in the fact that the map may be implicitly defined by a property or must be extrapolated from prototypes. An entropy measure and an equivocation measure are defined that permit an assessment of the improvement gained (and the price in confusion paid) by a set of Linear ``features'' are identified as measures and L 2 functions, respectively. It is shown that certain important normalizations (position, size, pitch, etc.) are nonlinear operations. Finally, the method of spectral analysis which is widely used for speech analysis is examined critically. It is shown that contrary to common belief Fourier analysis is not very suitable for detecting certain speech particles (consonants, stops, etc.).

19 citations


Proceedings ArticleDOI
H. Ryan1
01 Dec 1968
TL;DR: The use of performance measures for selecting features in pattern recognition systems is reviewed and an approximation to the information content measure is derived that reduces the computation required to calculate the measure.
Abstract: The use of performance measures for selecting features in pattern recognition systems is reviewed. An approximation to the information content measure is derived. The approximation reduces the computation required to calculate the measure. The accuracy of the approximation depends directly on the nature of the patterns and their features. Computational requirements for the approximate measure are specified.

15 citations


Patent
28 May 1968
TL;DR: In this article, a means of electrically extracting optical information from an optically recognizable pattern regardless of the orientation of the pattern within the field of view and comparing the information extracted against electrical criteria so as to classify the pattern was proposed.
Abstract: A means of electrically extracting optical information from an optically recognizable pattern regardless of the orientation of the pattern within the field of view and comparing the information extracted against electrical criteria so as to classify the pattern wherein the pattern image is projected onto an image slicer which utilizes fiber optics to divide the image into a plurality of slices. Electrical processing is used to assign a number to each slice which is proportional to the light flux incident upon the slice and to square the assigned number. All the squared numbers are added to generate a voltage which is memorized. The image is then rotated with respect to the slicer and subsequent voltages are generated and memorized. The memorized voltages are then compared to an electrical template to determine the classification of the pattern and its orientation within the field of view.

13 citations


Journal ArticleDOI
01 Dec 1968
TL;DR: State variable techniques are utilized here to yield efficient computer-implementable procedures for obtaining the double orthogonal expansion of the observable random process under hypothesis H i, i = 1, 2.
Abstract: Optimal continuous linear feature extraction for the binary Gaussian pattern recognition or detection problem necessitates finding the double orthogonal expansion of the observable random process under hypothesis H i , i = 1, 2. State variable techniques are utilized here to yield efficient computer-implementable procedures for obtaining the double orthogonal expansion.

11 citations


Journal ArticleDOI
TL;DR: In this paper, a direct-search computer program using a heuristic approach is described, making an attempt to extract feature informations automatically from patterns which may consist of open lines, partially overlapping cells, and cells that may lie entirely inside another cell.
Abstract: An attempt is made to extract feature informations automatically from patterns which may consist of open lines, partially overlapping cells, and cells that may lie entirely inside another cell. The usual pattern-recognition techniques, such as the linear threshold logic technique and the masking or template technique, are not practical here, if not entirely impossible. In this paper, a direct-search computer program using a heuristic approach is described. A test pattern is used to illustrate the capability of the program. The subject should be of general interest to those in the field of automation and cybernetics.

10 citations


Journal ArticleDOI
TL;DR: Three suboptimal solutions are obtained for the joint sequential feature selection and pattern classification problem and these solutions allow the comparison of two distinctly different approximations to the optimal procedure.
Abstract: —In this note, three suboptimal solutions are obtained for the joint sequential feature selection and pattern classification problem. These solutions allow the comparison of two distinctly different approximations to the optimal procedure. One approximation involves simplifying assumptions on the underlying distribution of features for each pattern class, while the second involves an approximation in the implementation of the optimal procedure.

7 citations


01 Jun 1968
TL;DR: A learning program designated CE, Concept-EPAM, is described that modifies EPAM through the introduction of a set membership relation and the effects of this extension are considered with respect to methods of storing concept descriptions in memory and methods of specifying learning and retrieval.
Abstract: : A learning program designated CE, Concept-EPAM, is described that modifies EPAM through the introduction of a set membership relation. The effects of this extension are considered with respect to methods of storing concept descriptions in memory and methods of specifying learning and retrieval. The learning strategies consist of interactions between image elaboration and tree modification. Implementations of CE are considered for a concept learning task and a pair associate task. Applicability of CE to a geometry analogy task requiring relational concepts is discussed. The relationship between the learning of concepts of concepts and feature extraction is illustrated.