Performance Of Pattern Recognition Methods Research Articles

The impact of the order of derivative spectra on the performance of pattern recognition methods. Classification of medicinal plants according to the phylum.

Data pre-processing is an important strategy in chemometrics and related fields because in many cases the transformation of data has a great effect on the performance of the method (model). However, a careful examination of the literature clearly points out that only very few systematic studies are dedicated to the effect of the derivative spectra on the performance of the pattern recognition methods. This comprehensive study compares the impact of the order of derivative spectra and other data pre-processing procedures (normalization and standardization) on the performance of cluster analysis, principal component analysis and discriminant analysis applied for characterization and classification of medicinal plants according to their phylum using UV spectra. The efficiency of the pre-processing methods was estimated by comparing the accuracy of classification and prediction measured by internal cross-validation. Derivatization method (1st order) resulted in the best classification (100%) of medicinal plants according to their phylum (Pteridophyte, Magnoliophyte and Spermatophyte) as compared to other pre-processing methods (normalized spectra-71.4%, standardized spectra-76.2% and original spectra-78.6%).

Extraction and Classification of Human Body Parameters for Gait Analysis

Human gait analysis is considered a new biometric tool for the ability to obtain metrics from the body at a distance. Biometric identifiers have properties that can technologically measure and analyze the characteristics of the human body and can be used as a form of identification and access control for security applications. Recognition through proper interpretation of gait parameters has become a relevant pattern classification problem. This work aims to develop an image processing system, with the use of the Microsoft Kinect sensor, which is capable to extract movement patterns for gait analysis and to present a comparative study of different pattern recognition methods for human identification. The image processing system, developed in C#, allowed the acquisition of three-dimensional data from several volunteers and made it possible to identify the human skeleton and automatically extract the kinetic and kinematic parameters of the body. For data analysis, different classification methods were compared. Among them, the algorithms that presented better performance were probabilistic neural networks, deep neural networks and k-nearest neighbors, with nearly 99% correct recognition rate. The obtained results demonstrate the efficiency of gait analysis as a biometric method. They also show the viability of gait parameter extraction using the Kinect sensor and the good performance of pattern recognition methods applied to the acquired gait kinetic and kinematic parameters.

Characterization of Cast Iron Microstructure Through Fluctuation and Fractal Analyses of Ultrasonic Backscattered Signals Combined with Classification Techniques

This work aims at evaluating the performance of pattern recognition methods in the identification of different microstructures presented by cast iron, namely, lamellar, vermicular and nodular microstructures, through the statistical fluctuation and fractal analyses of backscattered ultrasonic signals. The signals were obtained with a broad band ultrasonic probe with a central frequency of 5 MHz. The statistical fluctuations of the ultrasonic signals were analyzed by means of Hurst (RSA) and detrended-fluctuation analyses (DFA), and the fractal analyses were carried out by applying the minimal cover and box-counting techniques to the signals. The curves obtained from the statistical fluctuations and fractal analyses, as functions of the time window, were processed by using four pattern classification techniques, namely, principal-component analysis (PCA), Karhunen-Loève transformation (KLT), neural networks and Gaussian classifier. The best results were obtained by Karhunen-Loève expansion and neural networks, where an approximately 100% success rate has been reached for the classification of the different microstructures as well as for the training and the testing sets of events. The results presented correspond to an average taken over 100 randomly chosen sets of events. These results indicate that, within the techniques used, the Karhunen-Loève transformation and neural network associated with the statistical fluctuation analyses (RSA and DFA) are the best tools for the recognition of the different cast iron microstructures. It is worthwhile pointing out that the microstructure classification was made by using backscattering signals acquired during pulse echo ultrasonic nondestructive testing only. Therefore, that approach is a promising method for material characterization.

MEMS 기술로 제작된 가스 센서 어레이를 이용한 유해가스 분류를 위한 간단한 통계적 패턴인식방법의 구현

We have been implemented simple statistical pattern recognition methods for harmful gases classification using gas sensors array fabricated by MEMS (Micro Electro Mechanical System) technology. The performance of pattern recognition method as a gas classifier is highly dependent on the choice of pre-processing techniques for sensor and sensors array signals and optimal classification algorithms among the various classification techniques. We carried out pre-processing for each sensor's signal as well as sensors array signals to extract features for each gas. We adapted simple statistical pattern recognition algorithms, which were PCA (Principal Component Analysis) for visualization of patterns clustering and MLR (Multi-Linear Regression) for real-time system implementation, to classify harmful gases. Experimental results of adapted pattern recognition methods with pre-processing techniques have been shown good clustering performance and expected easy implementation for real-time sensing system.

Lateral filtering of medical ultrasonic B-scans before image generation.

A method is presented for improving the lateral resolution of diagnostic B-scan images. Wiener filtering is applied on sequences of A-scan echo amplitudes acquired during B-scanning, using a simple scanning action. The first applications to clinical B-scan data show an improvement in the image quality. The filtered images obtained by this techniques may provide an improvement in the performance of pattern recognition methods on medical ultrasonic image data.

Lateral filtering of medical ultrasonic B-scans before image generation

A method is presented for improving the lateral resolution of diagnostic B-scan images. Wiener filtering is applied on sequences of A-scan echo amplitudes acquired during B-scanning, using a simple scanning action. The first applications to clinical B-scan data show an improvement in the image quality. The filtered images obtained by this technique may provide an improvement in the performance of pattern recognition methods on medical ultrasonic image data.