BazTech - Yadda

1

Hybrid texture and gradient modeling for dynamic background subtraction identification systemin tobacco plant using 5G data service

Gowda Thirthe M.T., Chandrika J.

Computer Assisted Methods in Engineering and Science

|

2023

|

Vol. 30, no. 1

41--54

EN

Background: Detecting the plants as objects of interest in any vision-based input sequence is highly complex due to nonlinear background objects such as rocks, shadows,etc. Therefore, it is a difficult task and an emerging one with the development of precision agriculture systems. The nonlinear variations of pixel intensity with illuminationand other causes such as blurs and poor video quality also make the object detection taskchallenging. To detect the object of interest, background subtraction (BS) is widely usedin many plant disease identification systems, and its detection rate largely depends on thenumber of features used to suppress and isolate the foreground region and its sensitivitytoward image nonlinearity. Methodology: A hybrid invariant texture and color gradient-based approach is proposed to model the background for dynamic BS, and its performance is validated byvarious real-time video captures covering different kinds of complex backgrounds and various illumination changes. Based on the experimental results, a simple multimodal featureattribute, which includes several invariant texture measures and color attributes, yieldsfinite precision accuracy compared with other state-of-art detection methods. Experimental evaluation of two datasets shows that the new model achieves superior performanceover existing results in spectral-domain disease identification model. 5G assistance: After successful identification of tobacco plant and its analysis, the finalresults are stored in a cloud-assisted server as a database that allows all kinds of 5G servicessuch as IoT and edge computing terminals for data access with valid authentication fordetailed analysis and references.

2

Preprocessing large datasets using Gaussian mixture modelling to improve prediction accuracy of truck productivity at mine sites

Fan Chengkai, Zhang Na, Jiang Bei, Liu Wei Victor

Archives of Mining Sciences

|

2022

|

Vol. 67, no. 4

661--680

EN

The historical datasets at operating mine sites are usually large. Directly applying large datasets to build prediction models may lead to inaccurate results. To overcome the real-world challenges, this study aimed to handle these large datasets using Gaussian mixture modelling (GMM) for developing a novel and accurate prediction model of truck productivity. A large dataset of truck haulage collected at operating mine sites was clustered by GMM into three latent classes before the prediction model was built. The labels of these latent classes generated a latent variable. Two multiple linear regression (MLR) models were then constructed, including the ordinary-MLR (O-MLR) and the hybrid GMM-MLR models. The GMM-MLR model incorporated the observed input variables and a latent variable in the form of interaction terms. The O-MLR model was the baseline model and did not involve the latent variable. The GMM-MLR model performed considerably better than the O-MLR model in predicting truck productivity. The interaction terms quantitatively measured the differences in how the observed input variables affected truck productivity in three classes (high, medium, and low truck productivity). The haul distance was the most crucial input variable in the GMM-MLR model. This study provides new insights into handling massive amounts of data in truck haulage datasets and a more accurate prediction model for truck productivity.

3

Intelligent velocity picking and uncertainty analysis based on the Gaussian mixture model

Wang Xiaowei, Gao Yang, Chen Chang, Yuan Huan, Yuan Sanyi

Acta Geophysica

|

2022

|

Vol. 70, no 6

2659--2673

EN

The stacking velocity is often obtained manually. However, manually picking is inefficient and is easily affected by subjective factors such as the priori information and the experience of different processors. To enhance its objectivity, efficiency and consistency, we investigated an unsupervised clustering intelligent velocity picking method based on the Gaussian mixture model (GMM). This method can automatically pick the stacking velocity fast, and provide uncertainty analysis as a quality control. Combined with the geometry feature of energy clusters in velocity spectra, taking advantages of the geometric diversity of energy clusters, GMM can ft the energy clusters with different distributions more appropriately. Then, mean values of the final several submodels are located as the optimal velocity, and the multiples are avoided under the expert knowledge and geological rules. In addition, according to the covariance of submodels, we can derive the uncertainty analysis of the final time-velocity pairs, so as to indicate the reliability of picking velocity at different depths. Moreover, the automated interpreted velocity field is used for both normal moveout (NMO) correction and stacking. The comparison with the manual references is adopted to evaluate the quality of the unsupervised clustering intelligent velocity picking method. Both synthetic data and 3D field data have shown that the proposed unsupervised intelligent velocity picking method can not only achieve similar accuracy with manual results, but also get rid of multiples. Furthermore, compared with manual picking, it can significantly improve the efficiency and accuracy in identifying pore and cave structures, as well as indicating the uncertainty of time-velocity pairs by variance.

4

Fitting a Gaussian mixture model through the Gini index

López-Lobato Adriana Laura, Avendaño-Garrido Martha Lorena

International Journal of Applied Mathematics and Computer Science

|

2021

|

Vol. 31, no. 3

487--500

EN

A linear combination of Gaussian components is known as a Gaussian mixture model. It is widely used in data mining and pattern recognition. In this paper, we propose a method to estimate the parameters of the density function given by a Gaussian mixture model. Our proposal is based on the Gini index, a methodology to measure the inequality degree between two probability distributions, and consists in minimizing the Gini index between an empirical distribution for the data and a Gaussian mixture model. We will show several simulated examples and real data examples, observing some of the properties of the proposed method.

5

Segmentation of aggregate and asphalt in photographic images of pavements

Mejía Angela Marcela, Alzate Marco Aurelio, Reyes-Ortiz Oscar Javier

Engineering Transactions

|

2021

|

Vol. 69, iss. 1

19--42

EN

Particle size distribution of aggregate in asphalt pavements is used for determining important characteristics like stiffness, durability, fatigue resistance, etc. Unfortunately, measuring this distribution requires a sieving process that cannot be done directly on the already mixed pavement. The use of digital image processing could facilitate this measurement, for which it is important to classify aggregate from asphalt in the image. This classification is difficult even for humans and much more for classical image segmentation algorithms. In this paper, an expert committee approach was used, including classical adaptive Otsu, k-means vector quantization over a set of 8 principal components obtained from 26 features, and a Gaussian mixture model whose parameters are estimated through the expectation-maximization algorithm. A novel cellular automata approach is used to coordinate these expert opinions. Finally, a simple heuristic is used to reduce sub- and over-segmentation. The segmentation results are comparable to those obtained by a human expert, while the sieve size of the segmented images corresponds very well with that obtained from the sieving process, validating the proposed method of segmentation. The results show that with the digital imaging procedure it was possible to detect particles with a size of 100 m with 90% of success with respect to time-consuming manual techniques. In addition, with these results it is possible to establish the homogeneity of the sample and the distribution of the particles within the asphalt mixture.

6

Integration of hidden markov models in the automated speaker recognition system for critical use

Kovtun Vjatcheslav V., Yukhimchuk Maria S., Kisała Piotr, Abisheva Akmaral, Rakhmetullina Saule

Przegląd Elektrotechniczny

|

2019

|

R. 95, nr 4

176--180

EN

In this article, the author theoretically substantiated the possibility of integration of hidden Markov models (IHMM) in the structure of the automated speaker recognition system for critical use (ASRSCU) for analysis of speech information from a plurality of independent input channels, which allowed within the statistical conception of pattern recognition to combine the accuracy of the approximation of input signals inherent the apparatus of GMM models. The authors proposed a mathematical apparatus for the integration of hidden Markov models, which allows us to adequately describe the set of interacting processes in the Markov paradigm with the preservation of temporal, asymmetric conditional probabilities between the chains.

PL

W tym artykule autorzy teoretycznie uzasadnili możliwość integracji ukrytych modeli Markowa (IHMM) w strukturze zautomatyzowanego systemu rozpoznawania głosu osoby mówiącej do zastosowań krytycznych (ASRSCU) do analizy informacji o mowie z wielu niezależnych kanałów wejściowych, które dopuszczają wewnątrz statystyczna koncepcję rozpoznawania wzorców w celu połączenia dokładności aproksymacji sygnałów wejściowych z aparatem modeli GMM. Autorzy zaproponowali aparat matematyczny do integracji ukrytych modeli Markowa, który pozwala odpowiednio opisać zestaw oddziałujących procesów w paradygmacie Markowa z zachowaniem czasowych, asymetrycznych warunkowych prawdopodobieństw między łańcuchami.

7

The Role of Climatic Variables for Estimating Probability of Abundance of Tree Species

Antúnez P., Wehenkel C., López-Sánchez C. A., Hernández-Díaz J. C.

Polish Journal of Ecology

|

2017

|

Vol. 65, nr 3

324--338

EN

We analyzed the influence of climatic variables on the abundance of native tree species in 1,490 sampling plots systematically distributed in the Sierra Madre Occidental (state of Durango, Northwestern Mexico, 26°50′ and 22°17′N and 107°09′ and 102°30′W). We used the Weibull distribution and the finite Gaussian mixture model to study the climatic limits of 15 tree species in relation to seven variables thought to affect species abundance. We found that although they may occur in the same geographical region, some species display a wider range of ecological tolerance than others. Of the 15 species under study, only two (Quercus magnoliifolia and Q. arizonica) can be considered generalists in relation to some climatic variables, while the other 13 species behaved as specialists, implying a narrower range of distribution. The analytical techniques used enabled us to demarcate the zones in which the probability of abundance of each species is highest in relation to the climate variables considered. The findings could be used to help define climate for the 15 studied tree species of economic and ecological interest.

8

Identyfikacja głosowa w otwartym zbiorze mówców

Kamiński K., Dobrowolski A. P., Majda-Zdancewicz E.

Przegląd Elektrotechniczny

|

2015

|

R. 91, nr 10

206-210

PL

W artykule zaprezentowano wyniki badań systemu automatycznego rozpoznawania mówcy, przeprowadzane z wykorzystaniem komercyjnej bazy głosów TIMIT. Głównym celem badań było rozszerzenie funkcjonalności systemu rozpoznawania mówcy poprzez dodanie układu progowego, a tym samym umożliwienie identyfikacji w otwartym zbiorze mówców. Przedstawiono różne warianty zastosowanego układu progowego oraz dokonano próby wzbogacenia wektora cech dystynktywnych o różnicę częstotliwości podstawowej wyznaczanej dwiema różnymi metodami.

EN

In the article there are presented the test results of the automatic speaker recognition system, conducted while using the commercial voice basis TIMIT. The main purpose of the test was to extend the functionality of the speaker recognition system by adding the threshold based system, and consequently to enable the identification in the open set of speakers. There are presented different application variants of the threshold based system and there is an attempt to enrich the vector of distinctive features with the fundamental frequency difference determined with two different methods.

9

Parametryzacja sygnału mowy w algorytmach rozpoznawania mowy

Wojtuń J., Ośka J., Piotrowski Z., Bernat M.

Elektronika : konstrukcje, technologie, zastosowania

|

2015

|

Vol. 56, nr 2

34-39

PL

Historia systemów automatycznego rozpoznawania mowy ma już kilkadziesiąt lat. Pierwsze prace badawcze z tego zakresu pochodzą z lat 50. XX wieku (prace w laboratoriach Bella oraz MIT). Pomimo iż zagadnieniem tym zajmuje się wiele zespołów badawczych na całym świecie, problem automatycznego rozpoznawania mowy nie został definitywne rozwiązany. Dostępne systemy rozpoznawania mowy nadal charakteryzują się gorszą skutecznością w porównaniu do umiejętności człowieka. W artykule przedstawiono schemat systemu rozpoznawania mowy na przykładzie rozpoznawania izolowanych słów języka polskiego. Zaprezentowano szczegółowy opis wyznaczania cech dystynktywnych sygnału mowy w oparciu o współczynniki mel – cepstralne oraz cepstralne współczynniki liniowej predykcji. Przedstawiono wyniki skuteczności rozpoznawania poszczególnych fraz.

EN

The first research in automatic speech recognition systems dates back to the fifties of the 20th century (the works of Bell Labs and MIT). Although this issue has been treated by many research teams, the problem of automatic speech recognition has not been definitively resolved and remains open. Available voice recognition systems still have a poorer efficiency compared to human skills. This article presents a diagram of speech recognition system for isolated words of the Polish language. A detailed description of the determination of distinctive features of the speech signal is presented based on the mel-frequency cepstral coefficient and linear predictive cepstral coefficients. Efficiency results are also presented.

10

Ocena funkcjonalności systemu rozpoznawania mówcy dla zdegradowanej jakości sygnału głosowego

Kamiński K., Dobrowolski A.P., Majda E.

Przegląd Elektrotechniczny

|

2014

|

R. 90, nr 8

164-167

PL

W artykule przedstawiono wyniki badań automatycznego systemu rozpoznawania mówcy (ASR – ang. Automatic Speaker Recognition), przeprowadzonych na podstawie komercyjnej bazy głosów TIMIT. Badania prowadzone były pod kątem zastosowania ASR jako systemu automatycznego rozpoznawania rozmówcy telefonicznego. Przedstawiono również wpływ liczebności bazy głosów oraz stopień oddziaływania kompresji stratnej MP3 na skuteczność rozpoznawania mówcy.

EN

The article presents the results of tests of an automatic speaker recognition system (ASR) conducted on the basis of the TIMIT commercial voice database. The research was conducted with the aim of using ASR as a system for automatic recognition of telephone callers. The impact of the number of voices in the database and the effect of lossy MP3 compression on the effectiveness of speaker recognition has also been shown.

11

Computer-supported process of proteomic data profiling

Plechawska-Wójcik M.

Logistyka

|

2014

|

nr 6

8730--8739

EN

The paper presents the complete and complex process of proteomic data profiling obtained in the process of mass spectrometry. The main described procedure is based on analyzing Gaussian Mixture Model representation of the data. However, the process of mass spectrometry data analysis is more complex and consists of a number of procedures, like data preparation, data pre-processing including baseline correction, detection of outliers and noise removal. The mean spectrum calculated based on received data is modeled with GMM and decomposed using the Expectation-Maximization algorithm. In this process localization of the mean spectrum peaks is found. Those results are applied into each single spectrum in the dataset in the form of Gaussian mask. The result in a set of data ready for further statistical analysis. This procedure enable to perform proteomic data profiling using specific statistical methods.

PL

Artykuł przedstawia kompletny, złożony proces profilowania danych proteomicznych uzyskanych w badaniach spektrometrycznych. Podstawowa opisana procedura analizy jest oparta o analizę modelu reprezentującego dane mieszanin gaussowskich. Jednak proces analizy danych spektrometrycznych jest bardziej złożony i składa się z wielu procedur, takich jak przygotowanie danych, wstępne przetwarzanie danych, detekcji danych odstających oraz usuwania szumu. Widmo średnie obliczone na podstawie otrzymanych danych jest modelowane w oparciu o model mieszanin gaussowskich dekomponowany za pomocą algorytmu maksymalizacji oczekiwanej wartości. W tym procesie wyszukiwane są pozycje pików z widma średniego. Otrzymane wyniki są nakładane na każde pojedyncze widmo w postaci maski gaussowskiej. Rezultatem jest zestaw danych gotowych do dalszej analizy statystycznej. Procedura ta pozwala wykonać profilowanie danych proteomicznych przy użyciu określonych metod statystycznych.

12

System automatycznego rozpoznawania mówcy z wykorzystaniem techniki cepstralnej i modeli mieszanin gaussowskich

Kamiński K., Dobrowolski A.P.

Przegląd Elektrotechniczny

|

2013

|

R. 89, nr 9

83-93

PL

W niniejszym artykule zaprezentowano zaimplementowany w środowisku Matlab system automatycznego rozpoznawania mówcy, wykorzystujący do opisu głosu unikatowy wektor cech, tzw. „odcisk głosu” (VP – ang. Voice Print). System używa w procesie klasyfikacji tzw. modele mieszanin Gaussowskich (GMM – ang. Gaussian Mixture Model). W końcowej części artykułu przedstawione są badania skuteczności rozpoznawania mówców dla różnych wariantów systemu oraz w różnych konfiguracjach jego parametrów.

EN

The paper discusses the system of automatic speaker recognition, implemented in Matlab environment and using a unique vector of features, the so-called voice print (VP) for voice description. The system uses the so-called Gaussian Mixture Models (GMM) for the classification process. The final section of the paper presents the studies on the efficiency of speaker recognition for various system versions and for different system parameter configurations.

13

Performance comparison of artificial neural network and Gaussian mixture model in classifying hand motions by using sEMG signals

Baspinar U., Varol H. S., Senyurek V. Y.

Biocybernetics and Biomedical Engineering

|

2013

|

Vol. 33, no. 1

33--45

EN

In this study, a home-made four channel sEMG amplifier circuit was designed for measuring of sEMG signals. The measured sEMG signals were recorded on to a computer with help of a DAQ board. The recorded sEMG signals were filtered first with a high-pass filter and afterwards a wavelet based filtering was applied to remove unwanted noises. Before applying of the wavelet based filtering, it was first determined which wavelet type, threshold selection rule and threshold would be suitable for the denoising process. As a second step, the recorded and denoised signals’ features were extracted. For classification of motions 8 time domain and 2 frequency domain features were used individually and in combinations. Lastly, seven different motions were classified and their classification performances were compared. In this study, classification rates of ANN and GMM classifiers were compared as regards features.

14

Speech emotion recognition under white noise

Huang C., Chen G., Yu H., Bao Y., Zhao L.

Archives of Acoustics

|

2013

|

Vol. 38, No. 4

457--463

EN

Speaker‘s emotional states are recognized from speech signal with Additive white Gaussian noise (AWGN). The influence of white noise on a typical emotion recogniztion system is studied. The emotion classifier is implemented with Gaussian mixture model (GMM). A Chinese speech emotion database is used for training and testing, which includes nine emotion classes (e.g. happiness, sadness, anger, surprise, fear, anxiety, hesitation, confidence and neutral state). Two speech enhancement algorithms are introduced for improved emotion classification. In the experiments, the Gaussian mixture model is trained on the clean speech data, while tested under AWGN with various signal to noise ratios (SNRs). The emotion class model and the dimension space model are both adopted for the evaluation of the emotion recognition system. Regarding the emotion class model, the nine emotion classes are classified. Considering the dimension space model, the arousal dimension and the valence dimension are classified into positive regions or negative regions. The experimental results show that the speech enhancement algorithms constantly improve the performance of our emotion recognition system under various SNRs, and the positive emotions are more likely to be miss-classified as negative emotions under white noise environment.

15

Implementacja systemu identyfikacji słów izolowanych przy zastosowaniu Modeli Mikstur Gaussowskich

Kamiński K.

Zeszyty Naukowe Wydziału ETI Politechniki Gdańskiej. Technologie Informacyjne

|

2012

|

T. 20

111--119

PL

W referacie przedstawiony został sposób działania systemu identyfikacji słów izolowanych, który w etapie klasyfikacji wykorzystuje Modele Mikstur Gaussowskich. Referat zawiera również wyniki testów skuteczności omawianego systemu w rozpoznawaniu cyfr. System został zaimplementowany w środowisku Matlab. Kolejnym etapem pracy autora będzie implementacja powyższego systemu na zestawie uruchomieniowym DSK 6713.

EN

The paper presents a method of identifying isolated words, which uses at the classification stage Gaussian Mixture Models. The paper also concerns test results of effectiveness of the discussed system in numbers recognition. The system was implemented in Matlab environment. The next stage of developer work is to implement the aforementioned system on a runtime set DSK 6713.

16

Subscriber authentication using GMM and TMS320C6713DSP

Piotrowski Z., Wojtuń J., Kamiński K.

Przegląd Elektrotechniczny

|

2012

|

R. 88, nr 12a

127-130

EN

The article presents the theoretical basis for the implementation of Gaussian Mixture Models and implementation of a word recognition system on the basis of DSK TMS302C6713 DSP from Texas Instruments. The effectiveness of the algorithm based on Gaussian Mixture Model has been demonstrated. The system was developed as a software module for voice authentication of a subscriber in a Personal Trusted Terminal (PTT). The PIN of a subscriber is verified through an utterance in the Personal Trusted Terminal.

PL

W artykule zaprezentowano teoretyczne podstawy realizacji Modeli Mikstur Gausowskich oraz implementację systemu rozpoznawania słów z wykorzystaniem zastawu uruchomieniowego DSK TMS302C6713 DSP firmy Texas Instruments. Zobrazowano skuteczność działania algorytmu opartego na Modelach Mikstur Gausowskich. System został opracowany jako moduł programowy na potrzeby głosowego uwierzytelniania abonenta w Osobistym Zaufanym Terminalu (PTT). Poprzez wypowiedzenie głosem swojego PIN-u abonent jest weryfikowany w Osobistym Zaufanym Terminalu.

17

A Type Design Method of Large Hydraulic Turbine Based on The Gaussian Mixture Model

Lin L., Zhong S., Xie X.

Przegląd Elektrotechniczny

|

2012

|

R. 88, nr 9b

153-156

EN

Because of the complex fluid motion in hydraulic turbine and the imperfect design theory, the selection design of Large-scale hydraulic turbine is achieved based on the calculation and analysis on the synthetic characteristic curves, which is subjectivity and low efficiency. To solve this problem, the Gaussian mixture model is used to extract the geometric features from the synthetic characteristic curves so that the retrieval process of the model wheel can be achieved by these geometric features. The search model of the running area from the synthetic characteristic curves is build based on the contour curve similarity transformation method. In the paper, the Monte Carlo method is adopted to obtain the mean values of the synthetic characteristics in the running area so that the evaluation targets can be established by combining the mean values with the hydraulic turbine design experiences. Finally the validity of running area can be evaluated by the evaluation targets. The test results show that the accuracy and efficiency of the selection design of Large-scale hydraulic turbine can be improved by the proposed method.

PL

W artykule przedstawiono problemy projektowania dużych turbin hydraulicznych. Zaproponowano model matematyczny bazujący na mieszanym modelu Gaussa w celu wydobycia parametrów geometrycznych. Zaadaptowano metodę Monte Carlo do określania średnich wartości syntetycznych charakterystyk w obszarze pracy.

18

Speaker recognition based on the combination of GMM and SVDD

Zhou Y., Zhang X., Wang J., Gong Y., Zhou Y.

Przegląd Elektrotechniczny

|

2011

|

R. 87, nr 3

329-332

EN

Scare-level combination of subsystems can yield significant performance gains over individual subsystems in speaker recognition. A novel speaker verification method based on support vector data description (SVDD) is proposed to remedy the defect of Gaussian mixture model (GMM) to same extent, and then using the theory of multiple classifier systems (MCS),a new speaker recognition system based on the combination of GMM and SVDD is proposed. Experiments on TlMIT speech database show that the GMM-SVDD model fully utilizes the complementarities of GMM and SVDD to improve the performance obviously in speaker verification, closed-set speaker identification and speaker recognition.

PL

Zaproponowano nowa metodę rozpoznawania głosu bazującą na systemie SVDD jako alternatywę dla modelu GMM. Następnie wykorzystując teorię wielokrotnego systemu klasyfikacji MCS zaproponowano wykorzystanie połączenia systemów GMM i SVDD. Eksperymenty potwierdziły że nowy model GMM-SVOO umożliwia ulepszonę rozpoznawanie głosu.

19

Speaker identification based on Gaussian mixture model : experiments with Polish language utterances

Dąbrowska A., Drgas S., Cetnarowicz D., Chmielewska I.

Elektronika : konstrukcje, technologie, zastosowania

|

2008

|

Vol. 49, nr 4

29-33

EN

In this paper results of experiments with the prototype speaker recognition system based on Gaussian mixture model (GMM) and mel-cepstral coefficients (MFCCs) are presented for Polish Corpora database [4]. The minimum amount of data to train a reliable model and the minimum length of a signal to recognize speakers have been determined. Furthermore, the speaker discriminative properties of Polish phonemes have been investigated. The phonemes with the best speaker discriminative properties have been determined.

PL

Przedstawiono eksperymenty identyfikacji mówcy za pomocą prototypowego systemu rozpoznawania mowy na podstawie sumy rozkładów normalnych (GMM) i współczynników mel-cepstralnych, (MFCC), uzyskanych z wykorzystaniem polskojęzycznej bazy Corpora [4]. W eksperymentach zbadano minimalną ilość danych potrzebnych do wytrenowania wiarygodnego modelu oraz długość sygnału wymaganą do poprawnej klasyfikacji. Ponadto przebadano dyskryminacyjne właściwości polskich fonemów do identyfikacji mówcy. Wyodrębniono fonemy, które w największym stopniu przyczyniają się do poprawnego rozpoznawania.