• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.1
• /
• pp.77-83
• /
• 2013

In this paper, we suggested a method about the improvement of the voice recognition rate and carried out a study on it. In general, voices were detected by applying the most widely-used method, HMM (Hidden Markov Model) algorithm. Regarding the method of detecting voices, the zero crossing ratio was calculated based on the units of voices before the existence of data was identified. Regarding the method of recognizing voices, the patterns shown by the forms of voices were analyzed before they were compared to the patterns which had already been learned. According to the results of the experiment, in comparison with the recognition rate of 80% shown by the existing HMM algorithm, the suggested algorithm based on the recognition of the patterns shown by the forms of voices showed the recognition rate of 92%, reflecting the recognition rate improved by about 12% compared to the existing one.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.12
• /
• pp.4934-4949
• /
• 2015

The unstable characteristic values of communication signals along with the varying SNR (Signal Noise Ratio) environment make it difficult to identify the modulations of signals. Most of relevant literature revolves around signal recognition under stable SNR, and not applicable for signal recognition at varying SNR. To solve the problem, this research developed a novel communication signal recognition algorithm based on Holder coefficient and cloud theory. In this algorithm, the two-dimensional (2D) Holder coefficient characteristics of communication signals were firstly calculated, and then according to the distribution characteristics of Holder coefficient under varying SNR environment, the digital characteristics of cloud model such as expectation, entropy, and hyper entropy are calculated to constitute the three-dimensional (3D) digital cloud characteristics of Holder coefficient value, which aims to improve the recognition rate of the communication signals. Compared with traditional algorithms, the developed algorithm can describe the signals' features more accurately under varying SNR environment. The results from the numerical simulation show that the developed 3D feature extraction algorithm based on Holder coefficient cloud features performs better anti-noise ability, and the classifier based on interval gray relation theory can achieve a recognition rate up to 84.0%, even when the SNR varies from -17dB to -12dB.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.10
• /
• pp.797-802
• /
• 2016

This paper proposes a noble voice recognition method based on an adaptive MFCC and deep learning for embedded systems. To enhance the recognition ratio of the proposed voice recognizer, ambient noise mixed into the voice signal has to be eliminated. However, noise filtering processes, which may damage voice data, diminishes the recognition ratio. In this paper, a filter has been designed for the frequency range within a voice signal, and imposed weights are used to reduce data deterioration. In addition, a deep learning algorithm, which does not require a database in the recognition algorithm, has been adapted for embedded systems, which inherently require small amounts of memory. The experimental results suggest that the proposed deep learning algorithm and HMM voice recognizer, utilizing the proposed adaptive MFCC algorithm, perform better than conventional MFCC algorithms in its recognition ratio within a noisy environment.

• Journal of the Institute of Convergence Signal Processing
• /
• v.14 no.2
• /
• pp.90-94
• /
• 2013

Research for Manchu character digitalization is at an early stage. This paper proposes a preprocessing algorithm for Manchu character recognition. This algorithm improves the existing Hilditch thinning algorithm so that it corrects thinning error for Manchu characters. The existing algorithm separates the characters into the left-hand side and right-hand side, while our alogorithm uses the central point between the points that strokes exist when it classifies each of characters. The experimentation results show that this method is valid for thinning and classification of Manchu characters.

• Speech Sciences
• /
• v.9 no.4
• /
• pp.253-263
• /
• 2002

In this paper, we propose new algorithm of speaker recognition which identifies the speaker using the information obtained by the intensive speech feature analysis such as pitch, intensity, duration, and formant, which are crucial parameters of individual voice, for candidates of high percentage of wrong recognition in the existing speaker recognition algorithm. For testing the power of discrimination of individual parameter, DTW (Dynamic Time Warping) is used. We newly set the range of threshold which affects the power of discrimination in speech verification such that the candidates in the new range of threshold are finally discriminated in the next stage of sound parameter analysis. In the speaker verification test by using voice DB which consists of secret words of 25 males and 25 females of 8 kHz 16 bit, the algorithm we propose shows about 1% of performance improvement to the existing algorithm.

• Phonetics and Speech Sciences
• /
• v.2 no.4
• /
• pp.137-143
• /
• 2010

In this paper, a speech endpoint detection algorithm is proposed. The proposed algorithm is a kind of state transition-based ones for speech detection. To reject short-duration acoustic pulses which can be considered noises, it utilizes duration information of all detected pulses. For the optimization of parameters related with pulse lengths and energy threshold to detect speech intervals, an exhaustive search scheme is adopted while speech recognition rates are used as its performance index. Experimental results show that the proposed algorithm outperforms the baseline state-based endpoint detection algorithm. At 5 dB input SNR for the beamforming input, the word recognition accuracies of its outputs were 78.5% for human voice noises and 81.1% for music noises.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.2
• /
• pp.757-770
• /
• 2020

We propose a hand gesture recognition method that is compatible with a head-up display (HUD) including small processing resource. For fast link adaptation with HUD, it is necessary to rapidly process gesture recognition and send the minimum amount of driver hand gesture data from the wearable device. Therefore, we use a method that recognizes each hand gesture with an inertial measurement unit (IMU) sensor based on revised correlation matching. The method of gesture recognition is executed by calculating the correlation between every axis of the acquired data set. By classifying pre-defined gesture values and actions, the proposed method enables rapid recognition. Furthermore, we evaluate the performance of the algorithm, which can be implanted within wearable bands, requiring a minimal process load. The experimental results evaluated the feasibility and effectiveness of our decomposed correlation matching method. Furthermore, we tested the proposed algorithm to confirm the effectiveness of the system using pre-defined gestures of specific motions with a wearable platform device. The experimental results validated the feasibility and effectiveness of the proposed hand gesture recognition system. Despite being based on a very simple concept, the proposed algorithm showed good performance in recognition accuracy.

• Journal of Digital Convergence
• /
• v.13 no.10
• /
• pp.353-358
• /
• 2015

The Vocabulary Recognition System made by recognizing the standard vocabulary is seen as a decline of recognition when out of the standard or similar words. In this case, reconstructing the system in order to add or extend a range of vocabulary is a way to solve the problem. This paper propose configured Bhattacharyya algorithm standing by speech recognition learning model using the Bayesian methods which reflect parameter estimation upon the model configuration scalability. It is recognized corrected standard model based on a characteristic of the phoneme using the Bayesian methods for parameter estimation of the phoneme's data and Bhattacharyya algorithm for a similar model. By Bhattacharyya algorithm to configure recognition model evaluates a recognition performance. The result of applying the proposed method is showed a recognition rate of 97.3% and a learning curve of 1.2 seconds.

• The Journal of the Korea Contents Association
• /
• v.13 no.1
• /
• pp.1-8
• /
• 2013

The accuracy of a real facial recognition system can be varied according to the accuracy of the eye detection algorithm when we design and implement a semi-automatic facial recognition algorithm depending on the eye position of a database. In this paper, a fully automatic facial recognition algorithm is proposed such that Gabor and LBP features are extracted from fiducial points of a face graph which was created by using fiducial points based on the eyes, nose, mouth and border lines of a face, fitted on the face image. In this algorithm, the recognition performance could be increased because a face graph can be fitted on a face image automatically and fiducial points based LPB features are implemented with the basic Gabor features. The simulation results show that the proposed algorithm can be used in real-time recognition for more than 1,000 faces and produce good recognition performance for each data set.

• The KIPS Transactions:PartB
• /
• v.12B no.5 s.101
• /
• pp.577-586
• /
• 2005

Many algorithms have been proposed for (ace recognition that is one of the most successful applications in image processing, pattern recognition and computer vision fields. Research for what kind of attribute of face that make harder or easier recognizing the target is going on recently. In flus paper, we propose method to improve recognition performance using relevance of face data and applied algorithms, because recognition performance of each algorithm according to facial attribute(illumination and expression) is change. In the experiment, we use n-tuple classifier, PCA and Gabor wavelet as recognition algorithm. And we propose three vectorization methods. First of all, we estimate the fitnesses of three recognition algorithms about each cluster after clustering the test data using k-means algorithm then we compose new clusters by integrating clusters that select same algorithm. We estimate similarity about a new cluster of test data and then we recognize the target using the nearest cluster. As a result, we can observe that the recognition performance has improved than the performance by a single algorithm without clustering.