Statistical Pattern Recognition Techniques Research Articles

Statistical pattern recognition techniques: A sample problem of ultrasonic determination of interfacial weakness in adhesive joints

A pattern recognition approach was applied to the analysis of ultrasonic echo signals from two classes of aluminum-to-aluminum adhesive bonds. The two classes differed in the surface preparation of the adherends prior to bonding, resulting in different interfacial properties of the joints. These properties have a crucial effect on the long-term adhesive properties of the specimens. Application of advanced signal processing and pattern recognition techniques enabled the classification of the joints according to the surface preparation of the adherends, based on features extracted from the ultrasonic signals. The statistics yielded an upper bound for the probability of mis-classification of the specimens. The sensitivity of certain features, extracted from the ultrasonic signal, to the interfacial characteristics of the specimens is explained by means of the natural frequencies of a joint's components and surface condition of the adherends. This leads to a method for selecting the optimal probe frequency for carrying out the ultrasonic inspection.

SOURCE IDENTIFICATION OF AN OIL SPILL ON TRINIDAD'S EAST COAST

ABSTRACT In April 1986, a large quantity of unrefined crude oil was released into the nearshore marine environment on Trinidad's east coast. The oil was observed to have affected approximately 20 kilometers of coastline. Physical examination of the oil collected revealed that it had the appearance of a light grade crude, was golden brown in color, and had a characteristic gassy odor. Also observed was a mass mortality of the bivalve Donax sp (chip-chip), which inhabits sandy areas of the intertidal zone. Oil was extracted from samples of water, beach sand, and chip-chip collected from selected stations along the affected area. Also collected were samples of oil from possible sources located both offshore and at land-based facilities. Chemical characterization of the oil extracts, using capillary gas chromatography, atomic absorption and fluorescence spectroscopy, and carbon-13 nuclear magnetic resonance spectrometry, yielded data sets that distinguished the oils in the environment from the suspect source oils investigated. Application of statistical pattern recognition techniques, utilizing a hierarchical clustering procedure, to data from both environmental and suspect source samples yielded a high probability match between the spill samples and samples from one of the two oil companies operating in the area.

Automatic detection of inter-frame motion in echocardiographic images.

Signal averaging of echocardiographic frames acquired over multiple cardiac cycles has been used to improve image quality. However, misalignment of frames from respiration, transducer or patient movement, and irregular cardiac contraction affects the quality of the resultant averaged image. A motion detection system has been developed using inter-frame subtraction and statistical pattern recognition techniques. Automatic selection of frames exhibiting significant motion in a canine cardiac model compare favorably to manual selection by cardiologists (phi = 0.94) in a test set of 103 images. This method, combined with signal averaging, has resulted in an improvement in image quality.

Classification and detection of single evoked brain potentials using time-frequency amplitude features.

The classification and detection of event-related brain potentials was investigated using signal processing and statistical pattern recognition techniques. Amplitudes at sampled time points and frequency quantities have previously been used as features. Improvements to these procedures were obtained by using features from the time-frequency plane to utilize the geometric relationship between time and frequency, capitalizing on the nonstationarity of the evoked potential signals. These features were transformed from the original data sets based upon a two-step classification/feature selection procedure which uses selected frequencies from step 1 as parameters for data filtering in step 2. Features were selected from the filtered data, classifiers were designed, and the estimated classification accuracies were computed.

Pattern recognition methods for optimizing multivariate tissue signatures in diagnostic ultrasound

Described is a supervised parametric approach to the detection and classification of disease from acoustic data. Statistical pattern recognition techniques are implemented to design the best ultrasonic tissue signature from a set of measurements and for a given task, and to rate its performance in a way that can be compared with other diagnostic tools. In this paper, we considered combinations of four ultrasonic tissue parameters to discriminate, in vivo , between normal liver and chronic active hepatitis. The separation between normal and diseased samples was made by application of the Bayes decision rule for minimum risk which includes the prior probability for the presence of disease and the cost of misclassification. Large differences in classification performance of various tissue parameter combinations were demonstrated using the Hotelling trace criterion (HTC) and receiver operating characteristic (ROC) analysis. The ability of additional measurements to increase or decrease discrimin-ability, even measurements from other diagnostic modalities, can be evaluated directly in this manner.

Pattern Recognition Methods for Optimizing Multivariate Tissue Signatures in Diagnostic Ultrasound

Described is a supervised parametric approach to the detection and classification of disease from acoustic data. Statistical pattern recognition techniques are implemented to design the best ultrasonic tissue signature from a set of measurements and for a given task, and to rate its performance in a way that can be compared with other diagnostic tools. In this paper, we considered combinations of four ultrasonic tissue parameters to discriminate, in vivo, between normal liver and chronic active hepatitis. The separation between normal and diseased samples was made by application of the Bayes decision rule for minimum risk which includes the prior probability for the presence of disease and the cost of misclassification. Large differences in classification performance of various tissue parameter combinations were demonstrated using the Hotelling trace criterion (HTC) and receiver operating characteristic (ROC) analysis. The ability of additional measurements to increase or decrease discriminability, even measurements from other diagnostic modalities, can be evaluated directly in this manner.

Spectral Pattern Recognition

The growth of interest in spectral pattern recognition has been stimulated by the need to build instruments that respond to one object or type of object, possibly amid a host of other objects. The needed instruments include analyzers or alarms, medical diagnostic devices such as electroencephalogram or electrocardiogram analyzers, process control devices, and the like. The extra information provided by multivariate (spectral) data is always useful, if not essential, for such difficult tasks. However, the quantity of such data that can be provided by modern instruments requires that statistical pattern recognition techniques be used to extract the useful information from the noise.

Trends in Pattern Recognition and Machine Learning

This paper is tutorial in nature introducing the statistical and syntactic pattern recognition technique. The problem of pattern recognition has special reference with image analysis and some aspects of modern methods and application of the area of shape analysis and detection of objects included.

Breast tissue classification using diagnostic ultrasound and pattern recognition techniques: I. Methods of pattern recognition

This paper discusses the application of statistical pattern recognition techniques to problems in diagnostic ultrasound. Using our own system as an example, we describe the concepts and specific methods that we have applied to a problem involving the computer-aided classification of breast tissue in vivo. Topics include feature generation, feature selection and classification, as well as a method which estimates the probability of error on classifying future data. An accompanying paper applies these methods to the classification of backscattered RF signals from normal and diseased breast tissue.

On-line diagnosis algorithm based on noise analysis

A minicomputer-based on-line diagnosis algorithm was developed and its verification test was performed. This algorithm identifies the plant status by applying a statistical pattern recognition technique to various multidimensional noise pattern vectors. In case the reference noise patterns are prepared for such abnormal plant states that the noise response characteristics are linear to different abnormality severeness values, this algorithm can discriminate the category and relative severeness of abnormality, irrespective of its severeness. The test was performed on-line with a simple analog simulator which was modeled on the experimental fast reactor JOYO. Seven pattern vectors were used to diagnose three categories of abnormality including abnormal control rod vibration. Reference patterns were learned under only one degree of severeness for each abnormality category. Test results showed the effectiveness of the proposed algorithm. This indicates that the number of reference patterns to be prepared can be significantly decreased.

The application of statistical pattern-recognition techniques in dating ancient architectural construction

This paper explores the application of statistical pattern-recognition methods, discriminant analysis in particular, to the problems of classifying and of dating remains of Byzantine brick construction in Constantinople and Asia Minor. The technique is introduced and compared to other relevant multivariate statistical methods; results of analysis are presented and discussed; substantive conclusions are drawn; and the strengths and weaknesses of the approach are critically discussed.

Pattern recognition of canine duodenal contractile activity.

An application of statistical pattern recognition technique to the classification of canine duodenal contractile activity resulting from the ingestion of three different test meals, viz. water, acid, and solid, is described. Feature vectors were extracted from the contractile activities recorded by means of strain gauges, and were based on average duration, number, and intensities of contractions, as well as their frequency and propagational characteristics. Pattern training and classification were performed via the Bayes decision rule. The resuits indicate that the three different meals give rise to three different kinds of contractile activities. An improvement in the classifier's performance was observed when it was updated with correctly classified patterns. The results appear to suggest that pattern analysis techniques add a new dimension to the study of gastrointestinal motility in health and disease. For example, pattern recognition techniques can also be extended to identify contractile patterns altered due to such pathological conditions as irritable bowel syndrome and postoperative diarrhea and, hence, elucidate the underlying relationship between contractile activity and distal propulsion of contents.

Two-Dimensional Hydrometeor Image Classification by Statistical Pattern Recognition Algorithms

Abstract The investigation reported here involves the automatic classification of binary (black and white) images of hydrometeors (ice particles and raindrops) taken from cloud samples. The goal is to classify such images (both complete and fractional) into the seven most common classes of hydrometeors by statistical pattern recognition techniques. Detailed investigation about the data acquisition system and preprocessing is made. Four moment invariants which yield good class separation were used as features for the classification process. A Bayes decision function which minimizes the probability of misclassification is used for classification. Bayes theorem is employed to update mean vectors and covariance matrices involved in the decision function. A discrete Kalman filtering algorithm is developed for the on-line estimation of the probability of occurrence of each class. For such estimation a discrete adaptive Kalman filtering algorithm is also developed which adjusts the filter gain matrix such as to ...

FOCAS - Faint Object Classification and Analysis System

view Abstract Citations (324) References (18) Co-Reads Similar Papers Volume Content Graphics Metrics Export Citation NASA/ADS Focas: faint object classification and analysis system. Jarvis, J. F. ; Tyson, J. A. Abstract The FOCAS (Faint Object Classification and Analysis System) apparatus for the detection and classification of images on astronomical plates which is intended to be used for the automatic assembly of catalogs of faint objects, is presented. The system has been used to process plates obtained by the KPNO 40-m telescope under good seeing conditions for 12 selected areas, which were digitized in 20-micron steps with a 20-sq micron pixel in a 6000 x 6000 raster corresponding to a field 37 arcmin on a side. The FOCAS acts to reduce the digitized image to a list including the magnitude, position and classification data for objects in the field as faint as the 24th magnitude, by a process including the segmentation of objects, object area assembly, and object assembly. Objects are then classified as noise, stars or galaxies on the basis of conventional nonparametric statistical pattern recognition techniques, and catalogs of objects meeting given criteria can be obtained by various programs. System tests have shown detection efficiencies of 97% and classification capabilities for 90 and 80% of the galaxies and stars, respectively, for apparent magnitudes down to 24, although the optimality and accuracy of the classifier have not been established. FOCAS has been used to assemble catalog of 86,000 objects which is 3 to 4 mag deeper than existing surveys. Publication: The Astronomical Journal Pub Date: March 1981 DOI: 10.1086/112907 Bibcode: 1981AJ.....86..476J Keywords: Astronomical Catalogs; Astronomical Photography; Celestial Bodies; Image Resolution; Systems Analysis; Classifications; Digital Systems; Galaxies; Magnitude; Pattern Recognition; Photographic Plates; Stars; Astronomy full text sources ADS | data products SIMBAD (7)

Application of a distance-based method for analysing spatial patterns to the analysis of radionuclide tomograms

The application of a statistical pattern recognition technique to the analysis of radionuclide tomograms of the liver was investigated. A diagnostic test of aggregation was applied to the spatial distribution of pixels within each colour band of the image and the behaviour of a resulting statistic used as a classifier for distinguishing five types of liver disease. The results of a preliminary training procedure are presented and show good separation of the classes.

Attributed Grammar-A Tool for Combining Syntactic and Statistical Approaches to Pattern Recognition

Attributed grammars are defned from the pattern recognidon point of view and shown to be useful for descriptions of syntactic stuctures as well as semantic attributes in primitives, subpatterns, and patterns. A pattern analysis system using attributed grammars Is proposed for pattern classification and description. This system extracts primitives and their attributes after preprocessing, performs syntax analysis of the resulting pattern representations, computes and extracts subpattern attributes for syntactically accepted patterns, and finally makes decisions according to the Bayes decision rule. Such a system uses a combination of syntactic and statistical pattern recognition techniques, as is demonstrated by illustrative examples and experimental results.

A Pattern Deformational Model and Bayes Error-Correcting Recognition System

A pattern deformational model is proposed in this paper. Pattern deformations are categorized into two types: local deformation and structural deformation. A structure-preserving local deformation can be decomposed into a syntactic deformation followed by a semantic deformation, the former being induced on primitive structures and the latter on primitive properties. Bayes error-correcting parsing algorithms are proposed accordingly which not only can perform normal syntax analysis but also can make statistical decisions. An optimum Bayes error-correcting recognition system is then formulated for pattern classification. The system can be considered as a hybrid pattern classifier which uses both syntactic and statistical pattern recognition techniques.

Pattern classes: A technique for recovering their distributions

Many statistical pattern-recognition techniques depend for their application on the generation of one or more prototype patterns for each decision class. In turn, the determination of prototypes is dependent on the underlying probability distribution associated with a given class and that distribution's relationship to the distributions associated with the remaining classes. If these distributions are known, the problem of classification is considerably less complex than if they are unknown. The problem of recovering an unknown underlying distribution is one that has received considerable attention. The results thus far, however, are nonpractical. A practical technique that makes use of certain parameters related to sample size is presented and verified.

On-line, adaptive speaker-independent word recognition system based on phonetic recognition techniques

The research reported in this paper deals with a new method of phonemic analysis of speech by statistical pattern recognition techniques and its application to the problem of Automatic Speech Recognition (ASR). An on-line, adaptive, trainable speaker-independent system is implemented using this approach. The details of the system follow: first, the beginning and end points of the speech utterance are detected. The utterance is then sent for automatic segmentation where it is segmented into the following classes: (1) voiced, (2) unvoiced, (3) transition, and (4) silence. An 11-dimensional feature vector consisting of 10 linear predictor coefficients and zero-crossing rate is extracted from these regions. For voiced and transition region, the feature extraction is done pitch synchronously and for unvoiced regions, a constant frame of 6.4 msec is used. A new phonetic unit called phoneme-pair is defined for the transition regions, while the unvoiced and voiced regions are represented using the phonemes of the IPA. Conditional probability densities for each of the phonemes and phoneme-pairs are estimated using non-parametric methods as a single polynomial in the 11-dimensional space. The classifier makes Bayes' minimum risk decision based on these probability densities. The recognition results of the ASR system are Training Set: 98.4%, Test Set: 96.0% (for speakers in the training set) and 91.0% (for speakers not in the training set). The present vocabulary of the system is 60 words and any new word can be added by entering its corresponding phonetic transcription. The adaptive and trainable characteristics of the system will also be demonstrated.

Experimental evaluation of information conveyed through voiced, unvoiced, and transition segments of speech

Studies in the perception of speech have so far been only of a qualitative nature using psychoacoustic experiments. In this paper a quantitative method of making statistical measures is described. The technique uses the concepts of Shannon's information theory and a measure of information content is formulated. An algorithm based on statistical pattern recognition techniques is used to segment speech into four different segment classes [K. Ganesan and W. C. Lin, “Statistical pattern recognition approach to speech segmentation,” J. Acoust. Soc. Am. 56, S31 (A) (1974)]. Spectrograms of each type of segment are then obtained through which the corresponding probability density functions are estimated. Let Cv, Ct, and Cu be the voiced, transition, and unvoiced segment classes respectively. Information conveyed through Cv is given by I(X|Cv) = ∑ x1,x2,x3 p(X) log p(X) − ∑ x1,x2,x3 p(X,Cv) log p(X|Cv), where X = {x1,x2,x3}, a vector point in the spectral space. Applying this formulation to all the three classes, we found that the information conveyed through Cv is significantly lower than that through Ct and Cu, which are approximately equal.