Showing papers on "Feature vector published in 1983"

Rapid discrimination of visual patterns

Abstract: Experiments involving the rapid discrimination of visual patterns are used to infer the spatial information available to an observer within the first few hundred ms of inspection. Eye movements are prevented by a very brief presentation of the stimulus, and the inspection interval is terminated by a presentation of a masking pattern. It is shown that detection of a single vertical target line segment, embedded in an array of differently oriented background segments, improves with the increase of mask delay. The reduction of the area in which the target may lie reduces the inspection time that is required to determine the target's presence or absence. The phenomena are invariant under changes of the spatial scale within the fovea and parafovea. These results are interpreted in the context of a model in which the diameter of the area which can be searched in parallel is proportional to the distance in a feature space between the target and background elements. The geometry of this feature space is similar to the functional architecture of the visual cortex. A theory of texture perception based on a qualitative all-or-none feature space of textons is described. A quantitative model that is shown to be essentially equivalent to the previous theory in the limiting case of very large feature differences is proposed.

Automatic classification of images

Abstract: A method and system for classifying a raster scanned target image by collapsing the image into a vector corresponding to a horizontal line of the image, mathematically transforming this vector into a feature vector which corresponds to the power spectrum of the collapsed image, and comparing the feature vector with stored decision rules corresponding to classes of possible target objects to determine if the image represents an object in one of the classes.

Speaker-independent word recognizer

Abstract: Speaker-independent word recognition is performed, based on a small acoustically distinct vocabulary, with minimal hardware requirements. After a simple preconditioning filter, the zero crossing intervals of the input speech are measured and sorted by duration, to provide a rough measure of the frequency distribution within each input frame. The distribution of zero crossing intervals is transformed into a binary feature vector, which is compared with each reference template using a modified Hamming distance measure. A dynamic time warping algorithm is used to permit recognition of various speaker rates, and to economize on the reference template storage requirements. A mask vector with each reference vector on a template is used to ignore insignificant (or speaker-dependent) features of the words detected.

Fast Transient Security Assessment

Abstract: The growth of large interconnected power systems demands a high degree of security for normal operation. This requirement emerges from the fact that the possible disturbances in large power systems could have catastrophic results. Therefore it is the duty of power system control center to retain the operating point within secure boundaries. In order to control the security of system, one has to monitor and analyze the level of system security and enhance it in case it does not fulfill the system security requirement. The pattern recognition technique which is proposed in this paper is a method suitable for fast, on-line security assessment. Most of the literature in this area concerns the use of steady state variables for identification pruposes; the use of transient measurements, however, has been found to be better suited for the fast classification of operating points with minimal error. One of the salient conclusions of this paper is that the pattern recognition technique is a serious candidate for on-line applications. This favorable conclusion is based on the inclusion of transient state variables in the feature vector. The main objective of the pattern recognition method in transient security assessment is to reduce the computational requirement to a minimum. This is done at the expense of elaborate off-line computations. The classical methodology of pattern recognition consists of defining a pattern vector x whose components consist of all significant variables of the system.

Identity verification system

Abstract: An input pattern of an individual requesting identity verification, such as a speech pattern, is compared with reference patterns of a plurality of categories in a category recognition unit (33) to recognize the category of the input pattern. A differential pattern corresponding to a difference between the input pattern and the reference pattern of the same category as that of the input pattern is computed by a differential pattern computation unit (35). The differential pattern is verified, in a verification unit, against reference differential patterns registered in a verification data file (31) to perform the identity verification. When the identity verification based on the differential pattern is difficult, a feature vector pattern distinctively representing a feature of the individual is extracted from the differential pattern. The feature vector pattern is verified, by the verification unit (341, against reference feature vector pattern data of the same category registered in the verification data file (31).

Pattern Recognition Experiments in the Mandala/Cosine Domain

Abstract: The problem of recognition of objects in images is investigated from the simultaneous viewpoints of image bandwidth compression and automatic target recognition. A scenario is suggested in which recognition is implemented on features in the block cosine transform domain which is useful for data compression as well. While most image frames would be processed by the automatic recognition algorithms in the compressed domain without need for image reconstruction, this still allows for visual image classification of targets with poor recognition rates (by human viewing at the receiving terminal). It has been found that the Mandala sorting of the block cosine domain results in a more effective domain for selecting target identification parameters. Useful features from this Mandala/cosine domain are developed based upon correlation parameters and homogeneity measures which appear to successfully discriminate between natural and man-made objects. The Bhattacharyya feature discriminator is used to provide a 10:1 compression of the feature space for implementation of simple statistical decision surfaces (Gaussian and minimum distance classification). Imagery sensed in the visible spectra with a resolution of approximately 5-10 ft is used to illustrate the success of the technique on targets such as ships to be separated from clouds. A data set of 38 images is used for experimental verification with typical classification results ranging from the high 80's to low 90 percentile regions depending on the options choosen.

Generalized chord transformation for distortion-invariant optical pattern recognition.

Abstract: An optical processor that realizes a generalized chord transformation is described. The wedge-ring detector samples of an autocorrelation are shown to be the histograms of the chord distributions. This dimensionality reduced set of features is used as the feature vector inputs for a Fisher linear classifier to determine the class of the input object independent of geometrical distortions. Initial discussions on the use of different classifiers, the polarity of the classifier’s output, and selection of the image training set are also advanced.

Variable frame length vocoder

Abstract: A variable frame length vocoder extracts a feature vector for each given frame, a predetermined number of frames being defined as a section. The feature vectors in each section are stored, changes in feature vectors within a section being approximated by a given number of variable time length flat sections with a constant time length portion between adjacent flat sections, adjacent flat sections being interconnected by an inclined section of the constant time length duration. A feature vector of each flat section is outputted as a representative vector of the flat section, and the number of frames comprising the flat section is outputted as a repeat signal. This information is processed at the synthesis side of the vocoder to produce the feature vector in each inclined section by interpolating the representative vectors of the flat sections on both sides of the inclined section.

Rule based strategies for image interpretation

Abstract: We present an interpretation system which utilizes world knowledge in the form of simple object hypothesis rules, and more complex interpretation strategies attached to object and scene schemata, to reduce the ambiguities in image measurements. These rules involve sets of partially redundant features each of which defines an area of feature space which represents a "vote" for an object. Convergent evidence from multiple interpretation strategies is organized by top-down control mechanisms in the context of a partial interpretation. One such strategy extends a kernel interpretation derived through the selection of object exemplars, which represent the most reliable image specific hypotheses of a general object class, resulting in the extension of partial interpretations from islands of reliability.

A doubly layered, genetic penetrance learning system

Abstract: The author's original state-space learning system (based on a probabilistic performance measure clustered in feature space) was effective in optimizing parameterized linear evaluation functions. However, more accurate probability estimates would allow stabilization in cases of strong feature interactions. To attain this accuracy and stability, a second level of learning is added, a genetic (parallel) algorithm which supervises multiple activations of the original system. This scheme is aided by the probability clusters themselves. These structures are intermediate between the detailed performance statistics and the more general heuristic, and they estimate an absolute quantity independently of one another. Consequently the system allows both credit localization at this mediating level of knowledge and feature interaction at the derived heuristic level. Early experimental results have been encouraging. As predicted by the analysis, stability is very good.

A learning system which accommodates feature interactions

Abstract: The author's state-space learning system has effectively optimized the coefficients of linear evaluation functions. The incremental approach uses statistical performance measures from completed solutions to bootstrap the heuristic, which estimates probability of task usefulness. These statistics are clustered in feature space, forming a mediating knowledge structure (region set) between the direct performance measures and the generalized evaluation function. The regions are data-determined, insensitive to noise, and allow management of interacting features through natural piecewise linearity. Early experiment with non linearity indicates stability, flexibility and improved task performance.

A comparison of three feature vector clustering procedures in a speech recognition paradigm

Abstract: One possible approach to achieving talker independence in discrete utterance recognition (DUR) is to classify speech feature vectors by using a talker-independent clustering procedure. There are many possible choices of clustering algorithms. This work studied the characteristics of three clustering procedures, Agglomerative, Basic Isodata, and a 'Biased Mean' modification of Basic Isodata, as applied to speech feature vectors. The feature extractor consisted of a six channel filterbank similar to those used in DUR systems. The speech data was derived from 19 (total) repetitions of a ten word vocabulary, spoken by 16 different talkers. Various distance functions and feature vector representations were employed. Agglomerative clustering did not produce clusters which corresponded to any apparent classification of speech events. The Biased Mean Isodata procedure did not converge, and therefore was not useful. The Basic Isodata algorithm produced clusters which were to varying degrees identifiable with classes of speech sounds. Simple classifiers for three such classes, based on these clusters, would classify feature vectors with 5-10% error rates. Best results were obtained by using feature vectors which consisted of the log filter channel energies. These test results are good enough to encourage further development of cluster-based feature vector classifiers.

A parametric approach to eddy current ndt signal processing

Abstract: An application involving the use of Fourier descriptors for characterizing eddy current impedance plane trajectories is described. The amplitude coefficients in the Fourier descriptors are incorporated into a feature vector along with the ‘phase angle” of the trajectory. Defect classification is obtained by clustering the feature vectors in the multidimensional feature space using the K-Means algorithm. Advantages of the method include the insensitivity of the method to drift in the gain and zero settings of the measuring eddy current instrument and also fluctuations in the probe speed. In addition the curve can be reconstructed using the descriptors. Experimental evidence showing that the method can be successfully used for identifying defect signatures is presented.

Pattern recognizing system

Abstract: PURPOSE:To recognize effectively a graphic pattern even if a noise exists in it, by expressing features of respective parts of the graphic pattern with a matrix and obtaining the degree of resemblance between each feature vector and standard patterns. CONSTITUTION:Feature information of the graphic pattern obtained by a feature quantity calculating part 2 is stored in a corresponding position of a pattern matrix buffer memory 4. In this memory 4, features in respective positions are constituted into the matrix with one of features and positions as rows and the other as columns, and the graphic pattern is expressed with this matrix. Every feature vector element is led to a composite resemblance degree calculating part 5, and dictionary patterns registered in a standard pattern matrix dictionary 6 are referred to, and the degree of resemblance is obtained by a composite resemblance degree calculation. This calculation result is given to a recognition result editing and outputting part 7, and the recognition result of the graphic pattern is obtained.

A Parallel Method for Natural Texture Synthesis.

Abstract: : This paper deals with an optimization technique applied to natural texture synthesis. It proposes a definition of a global criterion which is the mean square error between the statistical features of a natural original texture and those of an artificially generated one. A gradient algorithm is used to minimize this criterion. The statistical feature vector used was the autocorrelation function although this is by no means the only choice. The textures generated are very similar to the original ones. This method can be implemented in a highly parallel manner. (Author)

Evaluation of Fourier Descriptors for Target Recognition in Digital Imagery.

Abstract: : This report describes the mathematical processes involved in generating co-efficients from a binary image of an object. It includes fourier descriptor theory, the normalized fourier descriptor, description of system implementation with library generation, chain code format, normalized fourier descriptor calculation, feature vector format, feature reduction, unknown feature vector extraction, feature vector classification and results applied to target recognition. (Author)

Feature extracting and sorting method of character pattern

Abstract: PURPOSE:To extract stable feature with simple processing, by obtaining a variation pattern of numbers of runs in terms of horizontal and vertical directions of a character pattern and then extracting the variation pattern of number of runs of two types as the feature of the character pattern. CONSTITUTION:A reading circuit 12 give horizontal and vertical raster scans to the character pattern in a character pattern memory 10 and feeds the character pattern data to white/black variation detectors 14X and 14Y. The number of black runs on each scanning line in X and Y directions are counted by black run counters 16X and 16Y. The difference is obtained by differential detectors 20X and 20Y between the number of runs obtained by counters 16X and 16Y and the number of runs on the immediately preceding line which are held temporarily in registers 18X and 18Y. The number of runs on the present scanning line are set to registers 22X and 22Y, and the feature vectors obtained at the registers 22X and 22Y are fed to a comparator 24 to be compared with the contents of a feature dictionary memory 27 for sorting.

Machine recognition of partial shapes using shape vectors

Abstract: A system for the machine recognition of partial shapes is described. Shape analysis methods and their limitations are reviewed in context with the problem of machine recognition of partial shapes. The problem of defining the critical points for shapes and partial shapes with various degrees of curvature is considered. It is shown that the critical points derived using criteria based on curvature alone are insufficient to describe shapes represented by smooth curves. A new method of detecting critical points, called the Adaptive Line of Sight (ALS) method, is described. The ALS method exhibits superior performance over the methods based on curvature alone. The critical points determined by this method are based on a set of coordinate axes that are dependent on the shape itself. This guarantees that the critical points detected are independent of size, rotation, and displacement of the shape. The results of applying this new procedure to actual shapes are demonstrated and discussed. The vector concept of shape space is introduced, and the need for the independence of the size variable to the shape vector is demonstrated. The shape space is described in terms of its properties. Two theorems necessary for the machine recognition of partial shapes are stated and proved using shape space properties. The critical points are organized into structural units called feature vectors using the concept of Line of Sight of a Point. The feature vectors are concatenated to form a global shape vector. Shapes are compared feature by feature, using a syntactic technique which will point out if the two shapes are similar or not. Examples are given for actual shape data.

Linear Algebra Techniques For Pattern Recognition: Feature Extraction Case Studies

Abstract: Many linear algebra operations, matrix inversions, etc. are required in pattern recogni­ tion as well as in signal processing. In this paper, we concentrate on feature extraction pattern recognition techniques (specifically a chord distribution and a moment feature space). For these two case studies, we note the various linear algebra operations required in distortion- invariant pattern recognition. Systolic processors can easily perform all required linear algebra functions. 1. INTRODUCTION Linear algebra operations are required in many signal processing applications and these have been extensively discussed elsewhere in this volume. In this paper, I note that similar operations are also required in many pattern recognition and object identification applica­ tions. In this paper, specific attention is given to feature extraction or feature space based pattern recognition problems and to viable applications such as achieving object rec­ ognition in the face of geometrical distortions in the input image. The two feature extrac­ tion case studies considered are the use of a chord distribution and a moment feature space. Each of these results in considerably different linear algebra operations required on the object features to achieve the desired object identification. Sections 2 and 3 address the chord distribution feature space and Section 4 addresses the moment feature space case study.

Parametric-histogramming technique for gross segmentation of color images of natural scenes (processing)

Abstract: This thesis describes a new systematic method for gross segmentation of color images of natural scenes. It has been developed within the context of the human visual system and mathematical pattern recognition theory. The eventual goal of the research is to integrate these two contexts to extract the visually distinct segments of an image which have vital importance for higher-level analysis or interpretation. A novel computational (pattern recognition) technique, called parametric-histogramming, is proposed in accordance with the human color perception and the Fisher criterion. This technique detects and isolates the image clusters efficiently and correctly using the 1-D parametric histograms of the L*,H('o),C* cylindrical coordinates of the (L*,a*,b*) - uniform color space in an unsupervised operation mode. In order to obtain the features most useful for a particular image, a new statistical-structural feature extraction method is devised in the form of a reference feature library containing several files. The underlying files tend to model the fundamental characteristics of uniformity, isolation, boundary, identation, texture, shadow and highlight patterns according to the grouping property of human eye and the Julesz conjecture. The dynamic operation characteristic of the feature selection phase is a significant property of the method. This type of operation, called dynamic feature space construction, enables the algorithm to select a particular feature space from the set of feature spaces so that image clusters in the selected space are more tractable and reliable.