Application Of Pattern Recognition Techniques Research Articles

Application of pattern recognition techniques to qualitative electrochemical analysis Part V. Trial retrieval of the electrode system

The identification capability of the pattern-recognition techniques developed hitherto in our laboratory has been tested, on the basis of an elementary statistical method, with the electrode systems, Pb2+/DME and Tl+/DME system, whose half-wave potentials lie close together. It has been proved that 25 qualitative parameters characterizing these electrode systems can be obtained and utilized for the discrimination between them at a high recognition rate.

Application of pattern recognition techniques to qualitative electrochemical analysis: Part V. Trial retrieval of the electrode system

The identification capability of the pattern-recognition techniques developed hitherto in our laboratory has been tested, on the basis of an elementary statistical method, with the electrode systems, Pb 2+/DME and Tl +/DME system, whose half-wave potentials lie close together. It has been proved that 25 qualitative parameters characterizing these electrode systems can be obtained and utilized for the discrimination between them at a high recognition rate.

Application of pattern recognition techniques to breast cancer detection: Ultrasonic analysis of 100 pathologically confirmed tissue areas

Ultrasonic pulse-echo rf waveform analysis and selected pattern recognition methods were applied to classification of breast tissue. Emphasis was placed on the classification of solid tissue areas since fluid areas are easily identified by present B-scan techniques. Pattern recognition techniques such as the Fisher Linear Discriminant (FLD), Probability Density Function (PDF) curves, jackknife estimate and committee vote were used to construct and evaluate a two class algorithm, malignant versus benign tissue areas. A data base consisting of frequency domain features from 100 pathologically confirmed tissue areas from 87 patients were used to train the algorithm. Algorithm performance was acquired via the generalized jackknife procedure to significantly reduce the bias frequently encountered in algorithm evaluation. Estimated values of algorithm performance are sensitivity and specificity values of 96 percent and 68 percent, respectively.

Application of pattern recognition techniques to breast cancer detection: ultrasonic analysis of 100 pathologically confirmed tissue areas.

Ultrasonic pulse-echo rf waveform analysis and selected pattern recognition methods were applied to classification of breast tissue. Emphasis was placed on the classification of solid tissue areas since fluid areas are easily identified by present B-scan techniques. Pattern recognition techniques such as the Fisher Linear Discriminant (FLD), Probability Density Function (PDF) curves, jackknife estimate and committee vote were used to construct and evaluate a two class algorithm, malignant versus benign tissue areas. A data base consisting of frequency domain features from 100 pathologically confirmed tissue areas from 87 patients were used to train the algorithm. Algorithm performance was acquired via the generalized jackknife procedure to significantly reduce the bias frequently encountered in algorithm evaluation. Estimated values of algorithm performance are sensitivity and specificity values of 96 percent and 68 percent, respectively.

Assessment of oil contamination in the marine environment by pattern recognition analysis of paraffinic hydrocarbon content of mussels

Data on the paraffinic hydrocarbon content of mussels ( Mytilus edulis and M. califomianus) were analyzed statistically in an attempt to assess oil contamination of several aquatic areas of the greater Puget Sound basin in the State of Washington. The mussels were collected quarterly over a period of two years from ten locations. Fundamental statistical treatment of the data did not reveal meaningful trends. Application of pattern recognition techniques to the hydrocarbon data set allowed the mussel samples to be successfully classified into groups, from areas near known sources of potential pollution or from areas remote from known pollution. Several samples however, could not be grouped consistently as either from contaminated or uncontaminated sites. In addition, the mussels from a particular area could be separated as to season of sampling. Seasonal variation in levels of individual paraffinic hydrocarbons were greater than annual changes. The n-C 16 hydrocarbon concentration was the most useful of the 34 features studied for establishing possible oil contamination of the areas. The n-C 16 hydrocarbon content was also used successfully to detect oil spills and to define uncontaminated locations in Puget Sound.

Application of pattern recognition techniques to qualitative electrochemical analysis Part III. Data compression of non-linear static characteristics of the electrode process

By the two-dimensional fast Hadamard transform, non-linear characteristics of the electrode process have been compressed into a small matrix which can be fed to an automatic retrieval system or a learning machine.

Application of pattern recognition techniques to qualitative electrochemical analysis: Part III. Data compression of non-linear static characteristics of the electrode process

By the two-dimensional fast Hadamard transform, non-linear characteristics of the electrode process have been compressed into a small matrix which can be fed to an automatic retrieval system or a learning machine.

Application of pattern recognition techniques to qualitative electrochemical analysis Part II. On-line measurements of double-layer capacity by the superimposition method and adaptive compensation of charging current by the capacitance multiplier

Application of pattern recognition techniques to qualitative electrochemical analysis Part II. On-line measurements of double-layer capacity by the superimposition method and adaptive compensation of charging current by the capacitance multiplier

Application of pattern recognition techniques to qualitative electrochemical analysis: Part II. On-line measurements of double-layer capacity by the superimposition method and adaptive compensation of charging current by the capacitance multiplier

A random voltage signal with a larger amplitude for the faradaic electrolysis, on which a high-frequency small sinusoidal ac voltage signal is superimposed, has been used as a input signal. The response current of the ac voltage signal consists of the charging current. Gain values and phase values of the electrochemical double layer, which are obtained by comparison of input and output of the high-frequency signal, are located against the sample value of the electrode potential ( E) at each moment using a microcomputer, and the images of both E−C dl (gain) and E− (phase) patterns come out gradually in the memory of the microcomputer, where C dl is the double-layer capacity and θ is the phase difference. The gain values are used for the control of the capacitance multiplier, which becomes an accurate and real-time simulator of the practical electrochemical double layer at the different electrode potential. An adaptive and real-time compensation of the charging current has been achieved merely by the analog subtraction of the current flowing through the capacitance multiplier (only the low-frequency potentiostatic signal is applied to its input terminal) instantaneously from the low-frequency response current.

A Method for Automating the Visual Inspection of Printed Wiring Boards

The application of pattern recognition techniques to manufacturing processes is a rapidly developing technology. Automatic verification of the quality of printed wiring boards (PWB's) using pattern recognition techniques is one potential application in this field. Qualitatively, this problem is finding small, irregular features in an environment of complicated, but larger and well-defined geometric features. In addition to the basic pattern recognition task, stringent performance requirements, both for throughput and accuracy, must be met if actual production usage is expected. The method employed in this study is based on characterizing 5 × 5 or 7 × 7 element binary patterns derived from the class of PWB's being inspected as good or defective. A database of 80 512 × 512 element images of PWB's was constructed and used to determine the number of unique patterns and their rates of occurrence. The major experimental result of this study is that less than 500 of the possible (15/16)224 5 × 5 patterns are needed to describe all the border containing patterns in the 80 images. It is also apparent that more patterns would be required if the training database was larger. The small number of patterns needed to represent virtually all of the normal border patterns suggests a two-stage inspection strategy. In the first stage, each border pattern from the PWB being inspected is compared to a previously prepared list.

The Application of Pattern Recognition Techniques in the Analysis of Metacarpophalangeal Lengths

Pattern recognition techniques have been applied to the analysis of metacarpophalangeal lengths obtained from hand radiographs. A data base consisting of healthy subjects, and subjects exhibiting achondroplasia, Down's syndrome, and Turner's syndrome was studied. A classification technique was found which effectively separates these conditions. The technique is suitable for computer automation.

Application of pattern recognition techniques to qualitative electrochemical analysis Part I. Feature extraction of non-linear characteristics of the electrode process by the two-dimensional fast hadamard transform

The feature extraction on non-linear characteristics of the electrode process is described. A model consisting of a linear element and a static non-linear element (ZNL=zero-memory non-linear system) is used to acquire the representation of the electrode process, from which the features of the electrode process should be extracted. That is, the electrode process is assumed to be composed of two processes, a solution process which has a dynamic characteristic due to the accumulation of materials in solution, and a surface process which has a non-linear characteristic like a kind of Butler's relation, etc. Previous work showed that a linear diffusion model with and without pseudo-first order chemical reactions made it possible to obtain the representation of the former process as a cross-power-spectral pattern (linear transfer function) between the current (I) and the surface concentration (c). For the latter process, it should be pointed out that its non-linear characteristics are nearly static due to the absence of material accumulation along the normal direction to the surface at the electrode surface. On the basis of this static non-linear model, this paper shows that the surface process can be represented by a pattern of a static curved surface, E-I-c (E= electrode potential) obtained by use of the computational measurements of the surface concentration. A two-dimensional fast Hadamard transform achieved high information compression in the feature extraction of the observed E-I-c pattern.

Application of pattern recognition techniques to qualitative electrochemical analysis: Part I. Feature extraction of non-linear characteristics of the electrode process by the two-dimensional fast Hadamard transform

The feature extraction on non-linear characteristics of the electrode process is described. A model consisting of a linear element and a static non-linear element (ZNL=zero-memory non-linear system) is used to acquire the representation of the electrode process, from which the features of the electrode process should be extracted. That is, the electrode process is assumed to be composed of two processes, a solution process which has a dynamic characteristic due to the accumulation of materials in solution, and a surface process which has a non-linear characteristic like a kind of Butler's relation, etc. Previous work showed that a linear diffusion model with and without pseudo-first order chemical reactions made it possible to obtain the representation of the former process as a cross-power-spectral pattern (linear transfer function) between the current ( I) and the surface concentration ( c). For the latter process, it should be pointed out that its non-linear characteristics are nearly static due to the absence of material accumulation along the normal direction to the surface at the electrode surface. On the basis of this static non-linear model, this paper shows that the surface process can be represented by a pattern of a static curved surface, E- I- c ( E= electrode potential) obtained by use of the computational measurements of the surface concentration. A two-dimensional fast Hadamard transform achieved high information compression in the feature extraction of the observed E- I- c pattern.

Application of pattern recognition techniques to the interpretation of severely overlapped voltammetric data: studies with experimental data

Application of pattern recognition techniques to the interpretation of severely overlapped voltammetric data: studies with experimental data

Application of pattern recognition techniques to the interpretation of severely overlapped voltammetric data: theoretical studies

Application of pattern recognition techniques to the interpretation of severely overlapped voltammetric data: theoretical studies

Application of pattern recognition techniques to solids mixing—feature extraction

Learning algorithms implemented via computers are effective for identifying and classifying patterns in random environments. The fundamental concepts governing the processes of feature extraction are presented. Orthogonal transformations which utilize rotational transformations of the data vectors are described as techniques for feature extraction. These concepts and techniques are applied to identification of the homogeneity of a solids mixture which is sampled by two schemes, namely, spot sampling and contact number sampling. Results show that the features extracted from the mixture pattern by the orthogonal transformation can be used to characterize the homogeneity of the mixture.

Mathematical methods of feature selection in pattern recognition

In the 15 years of its existence pattern recognition has made considerable progress on both the theoretical and practical fronts. Starting from the original application of pattern recognition techniques to the problem of character recognition at the time when pattern recognition was conceived these techniques have now penetrated such diverse areas of science as medical diagnosis, remote sensing, finger prints and speech recognition, image classification, etc.* This wide applicability derives from the inherent generality of pattern recognition, which is a direct consequence of the adopted threestage concept of pattern recognition process. According to this concept the process of pattern recognition is viewed as a sequence of three independent functions--representation, feature selection and classification (Fig. 1). Among these functions only the representation stage, which transforms the input patterns into a form suitable for computer processing, is problemdependent. Both the feature selector, the function of which is to reduce the dimensionality of the representation vector, and the classifier, which carries out the actual decision process, work with a vector of measurements which can be considered as an abstract pattern. As a result, the feature selection and classification stages can be implemented using mathematical methods irrespective of the original application. Naturally, this has had a beneficial effect on the progress in the theory of pattern recognition. Although all three stages of the pattern recognition system play an essential role in the process of classifying patterns by machine, the quality of the system's performance depends chiefly on the feature selector. The reasons

Information processing of remotely sensed agricultural data

Systems utilizing sensors on aerospace platforms for the purpose of surveying earth resources have been under study recently. In this paper research on the information processing portion of such a system for agricultural resources is described. In particular the application of pattern recognition techniques to muItispectral data is proposed and example results are presented. Good classification accuracy on crop species identification of some 40 000 samples in a five square mile area was achieved in one example. Other examples show results in feature selection and nonparametric classifier algorithms.

Image processing in the biomedical sciences

A survey of current research and development activities in the automated analysis of biomedical images discusses the application of computer techniques to the extraction of information from photographs, x-ray films, and images on microscope slides. Accomplishments in the primary areas of endeavor include the analysis of chromosomes, the differentiation of white blood cells, the interpretation of radiographic images, and other areas of investigation. The limitations and problems encountered in the biomedical application of pattern recognition techniques are also discussed. Existing and potential value of utilizing automated image processing in biomedical investigations and medical practice are considered in terms of increasing analytic capability as well as simply quantifying biomedical data which occurs in visual form.