• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.2
• /
• pp.267-274
• /
• 2015

In this paper the optimal structure of a neural network based on OpenCV for a golf ball recognition and the intensity of ROI(Region Of Interest) are calculated. The system is composed of preprocess, image processing and machine learning, and a learning model is obtained by multi-layer perceptron using the inputs of 7 Hu's invariant moments, box ration extracted by vertical and horizontal length or ${\pi}$ calculated by area of ROI. Simulation results show that optimal numbers of hidden layer and the node of neuron are selected to 2 and 9 respectively considering the recognition rate and running time, and optimal intensity of ROI is selected to 200.

• Journal of KIISE:Software and Applications
• /
• v.31 no.4
• /
• pp.403-410
• /
• 2004

Bimodal speech recognition systems have been proposed for enhancing recognition rate of ASR under noisy environments. Visual feature extraction is very important to develop these systems. To extract visual features, it is necessary to detect exact lip position. This paper proposed the method that detects a lip position using color similarity model and SVM. Face/Lip color distribution is teamed and the initial lip position is found by using that. The exact lip position is detected by scanning neighbor area with SVM. By experiments, it is shown that this method detects lip position exactly and fast.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.7
• /
• pp.1881-1892
• /
• 1997

This paper proposes the method for recognition of printed chinese characters by probabilistic VQ networks and multi-stage recognizer has hierarchical structure. We use modular neural networks, because it is difficult to construct a large-scale neural network. Problems in this procedure are replaced by probabilistic neural network model. And, Confused Characters which have significant ratio of miss-classification are reclassified using the entropy theory. The experimental object consists of 4,619 chinese characters within the KSC5601 code except the same shape but different code. We have 99.33% recognition rate to the training data, and 92.83% to the test data. And, the recognition speed of system is 4-5 characters per second. Then, these results demonstrate the usefulness of our work.

• Smart Media Journal
• /
• v.5 no.3
• /
• pp.49-53
• /
• 2016

Related researches are progressing that method of non-contact method using the electromagnetic field on the human body by detecting the motion recognition signal is the limitations of time and space, so less than the existing systems. In this paper, we designed the circuit system that can implement the hardware that can detect the electric field signal of the human body non-contact method to increase the recognition rate to screen this digital waveform. The PCB design Used to automatically increase of composition of the circuit and the linkage of the comparator digital waveform with circuit simulation of the system. At same time for evaluate the characteristics of the whole circuit system.

• Journal of information and communication convergence engineering
• /
• v.2 no.2
• /
• pp.132-137
• /
• 2004

Most of the current computer vision theories are based on hypotheses that are difficult to apply to the real world, and they simply imitate a coarse form of the human visual system. As a result, they have not been showing satisfying results. In the human visual system, there is a mechanism that processes information due to memory degradation with time and limited storage space. Starting from research on the human visual system, this study analyzes a mechanism that processes input information when information is transferred from the retina to ganglion cells. In this study, a model for the characteristics of ganglion cells in the retina is proposed after considering the structure of the retina and the efficiency of storage space. The MNIST database of handwritten letters is used as data for this research, and ART2 and SOM as recognizers. The results of this study show that the proposed recognition model is not much different from the general recognition model in terms of recognition rate, but the efficiency of storage space can be improved by constructing a mechanism that processes input information.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.11
• /
• pp.2275-2280
• /
• 2009

u-Healthcare service is one of attractive applications in ubiquitous environment. In this paper, we propose a method to recognize exercises using a new accelerometer based body-attached platform for supporting u-Healthcare service. The platform consists of a device for measuring accelerometer data and a device for receiving the data. The former measures a user's motion data using a 3-axis accelerometer. The latter transmits the accelerometer data to a computer for recognizing the user's exercise. The algorithm for exercise recognition classifies the type of exercise using principle components analysis(PCA) from the accelerometer data transformed by discrete fourier transform(DFT), and estimates the repetition count of the recognized exercise using a peak detection algorithm. We evaluate the performance of the algorithm from the accuracy of the recognition of exercise type and the error rate of the estimation of repetition count. In our experimental result, the algorithm shows the accuracy about 98%.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.6 no.4
• /
• pp.291-295
• /
• 2005

In this paper, we propose a face recognition method using modified 2-D PCA. While the previous PCA method computes the covariance matrix by using one dimensional vectors, the 2-D PCA method computes the covariance matrix by directly using direct two dimensional image, and extracts the feature vectors by solving eigenvalue problem. The proposed method recognizes the faces by applying the modified 2-D PCA to face images and it gets linear transformation matrix using two covariance matrices. The experimental results indicates that the proposed method improved about $1\%$ and achieved more stability in recognition rate than conventional 2-D PCA.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.3
• /
• pp.416-421
• /
• 2008

This paper realizes a system that recognizes traffic safety signs by applying the principle used for game in reverse. The game used for this paper is one that expresses the shape of temporary objects intended by the maker when the maker sees the numerical image provided on (x, y) coordinates and then expresses it on the mesh. After separating the traffic safety sign image from the input image, the system is realized by outputting the content of the sign into letters by recognizing the forms and colors constituting the sign using the puzzle game above. Our system has fast process time and better rate of recognition than the existing system with black-and-white image processing and recognition without any penciling progress.

• Phonetics and Speech Sciences
• /
• v.10 no.4
• /
• pp.111-119
• /
• 2018

This paper proposes a method to improve speech recognition performance by extracting various pronunciations of the pseudo-morpheme unit from an eojeol unit corpus and generating a new recognition unit considering pronunciation variations. In the proposed method, we first align the pronunciation of the eojeol units and the pseudo-morpheme units, and then expand the pronunciation dictionary by extracting the new pronunciations of the pseudo-morpheme units at the pronunciation of the eojeol units. Then, we propose a new recognition unit that relies on pronunciation by tagging the obtained phoneme symbols according to the pseudo-morpheme units. The proposed units and their extended pronunciations are incorporated into the lexicon and language model of the speech recognizer. Experiments for performance evaluation are performed using the Korean speech recognizer with a trigram language model obtained by a 100 million pseudo-morpheme corpus and an acoustic model trained by a multi-genre broadcast speech data of 445 hours. The proposed method is shown to reduce the word error rate relatively by 13.8% in the news-genre evaluation data and by 4.5% in the total evaluation data.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.2
• /
• pp.771-789
• /
• 2019

Text detection has been a popular research topic in the field of computer vision. It is difficult for prevalent text detection algorithms to avoid the dependence on datasets. To overcome this problem, we proposed a novel unsupervised text detection algorithm inspired by bootstrap learning. Firstly, the text candidate in a novel form of superpixel is proposed to improve the text recall rate by image segmentation. Secondly, we propose a unique text sample selection model (TSSM) to extract text samples from the current image and eliminate database dependency. Specifically, to improve the precision of samples, we combine maximally stable extremal regions (MSERs) and the saliency map to generate sample reference maps with a double threshold scheme. Finally, a multiple kernel boosting method is developed to generate a strong text classifier by combining multiple single kernel SVMs based on the samples selected from TSSM. Experimental results on standard datasets demonstrate that our text detection method is robust to complex backgrounds and multilingual text and shows stable performance on different standard datasets.