• The Journal of the Acoustical Society of Korea
• /
• v.29 no.1
• /
• pp.56-61
• /
• 2010

This paper describes optimization methods of acoustic models in HMM-based continuous speech recognition. Most of the conventional speech recognition systems use the same number of Gaussian mixture components for each HMM state. However, since the number of data samples available for each state is different from each other, it is possible to reduce the overall number of model parameters and the computational cost at the decoding step by optimizing the number of Gaussian mixture components. In this study, we introduced the Gaussian mixture weight term at the merging stage of Gaussian components in the minimum description length (MDL) based acoustic modeling optimization. Experimental results showed that the proposed method can obtain better ASR accuracy than the previous optimization method which does not consider the Gaussian mixture weight term.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.4
• /
• pp.2038-2057
• /
• 2017

Gaussian learning is a new technology in the computational intelligence area. However, this technology weakens the learning ability of a particle swarm and achieves a lack of diversity. Thus, this paper proposes a vector Gaussian learning strategy and presents an effective approach, named particle swarm optimization based on vector Gaussian learning. The experiments show that the algorithm is more close to the optimal solution and the better search efficiency after we use vector Gaussian learning strategy. The strategy adopts vector Gaussian learning to generate the Gaussian solution of a swarm's optimal location, increases the learning ability of the swarm's optimal location, and maintains the diversity of the swarm. The method divides the states into normal and premature states by analyzing the state threshold of the swarm. If the swarm is in the premature category, the algorithm adopts an inertia weight strategy that decreases linearly in addition to vector Gaussian learning; otherwise, it uses a fixed inertia weight strategy. Experiments are conducted on eight well-known benchmark functions to verify the performance of the new approach. The results demonstrate promising performance of the new method in terms of convergence velocity and precision, with an improved ability to escape from a local optimum.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.3
• /
• pp.629-635
• /
• 2017

In this paper, we proposed a nonlinear synthesis filter for noise reduction to reduce the effects of Gaussian noise and impulse noise. When the centralization of the local mask is judged to be Gaussian noise by the noise judgment, the weight value of the weight filter are applied differently according to the spatial weight filter and the pixel change by using the sample variance in the local mask. And if it is determined as the impulse noise, we proposed an algorithm that applies different weights of local histogram weight filter and standard median filter according to noise density of mask. In order to evaluate the performance of the proposed filter algorithm, we used PSNR(peak signal to noise ratio) and compared existing methods and proposed filter algorithm in the mixed noise environment with Gaussian noise, impulsive noise, and two noises mixed.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.11a
• /
• pp.73-76
• /
• 2018

A Gaussian noise is caused by surrounding environment or channel interference when transmitting image. The noise reduces not only image quality degradation but also high-level image processing performance. The Non-Local Means (NLM) filter finds similarity in the neighboring sets of pixels to remove noise and assigns weights according to similarity. The weighted average is calculated based on the weight. The NLM filter method shows low noise cancellation performance and high complexity in the process of finding the similarity using weight allocation and neighbor set. In order to solve these problems, we propose an algorithm that shows an excellent noise reduction performance by using Summed Square Image (SSI) to reduce the complexity and applying the weighting function based on a cosine Gaussian kernel function. Experimental results demonstrate the effectiveness of the proposed algorithm.

• Journal of information and communication convergence engineering
• /
• v.19 no.1
• /
• pp.54-60
• /
• 2021

Various digital devices were supplied throughout the Fourth Industrial Revolution. Accordingly, the importance of data processing has increased. Data processing significantly affects equipment reliability. Thus, the importance of data processing has increased, and various studies have been conducted on this topic. This study proposes a modified Gaussian filter algorithm based on a fuzzy membership function. The proposed algorithm calculates the Gaussian filter weight considering the standard deviation of the filtering mask and computes an estimate according to the fuzzy membership function. The final output is calculated by adding or subtracting the Gaussian filter output and estimate. To evaluate the proposed algorithm, simulations were conducted using existing additive white Gaussian noise removal algorithms. The proposed algorithm was then analyzed by comparing the peak signal-to-noise ratio and differential image. The simulation results show that the proposed algorithm has superior noise reduction performance and improved performance compared to the existing method.

• Smart Media Journal
• /
• v.4 no.3
• /
• pp.50-55
• /
• 2015

In this paper, we present background modeling method based on Gaussian mixture model to subtract background for night-time video surveillance. In night-time video, it is hard work to distinguish the object from the background because a background pixel is similar to a object pixel. To solve this problem, we change the pixel of input frame to more advantageous value to make the Gaussian mixture model using scaled histogram stretching in preprocessing step. Using scaled pixel value of input frame, we then exploit GMM to find the ideal background pixelwisely. In case that the pixel of next frame is not included in any Gaussian, the matching test in old GMM method ignores the information of stored background by eliminating the Gaussian distribution with low weight. Therefore we consider the stacked data by applying the difference between the old mean and new pixel intensity to new mean instead of removing the Gaussian with low weight. Some experiments demonstrate that the proposed background modeling method shows the superiority of our algorithm effectively.

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.4E
• /
• pp.3-10
• /
• 1998

This paper concerns the dynamical behavior, in probabilistic sense, of a feedforward neural network performing auto association for novelty. Networks of retinotopic topology having a one-to-one correspondence between and output units can be readily trained using back-propagation algorithm, to perform autoassociative mappings. A novelty filter is obtained by subtracting the network output from the input vector. Then the presentation of a "familiar" pattern tends to evoke a null response ; but any anomalous component is enhanced. Such a behavior exhibits a promising feature for enhancement of weak signals in additive noise. As an analysis of the novelty filtering, this paper shows that the probability density function of the weigh converges to Gaussian when the input time series is statistically characterized by nonsymmetrical probability density functions. After output units are locally linearized, the recursive relation for updating the weight of the neural network is converted into a first-order random differential equation. Based on this equation it is shown that the probability density function of the weight satisfies the Fokker-Planck equation. By solving the Fokker-Planck equation, it is found that the weight is Gaussian distributed with time dependent mean and variance.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.9
• /
• pp.1176-1182
• /
• 2021

Recently, with the development of artificial intelligence and IoT technology, automation, and unmanned work are progressing in various fields, and the importance of image processing, which is the basis of AI object recognition, is increasing. In particular, in systems that require detailed data processing, noise removal is used as a preprocessing step, but the existing algorithm does not consider the noise level of the image, so it has the disadvantage of blurring in the filtering process. Therefore, in this paper, we propose a modified Gaussian filter that determines the weight by determining the noise level of the image. The proposed algorithm obtains the noise estimate for the AWGN of the image using quadtree segmentation, determines the Gaussian weight and the pixel weight, and obtains the final output by convolution with the local mask. To evaluate the proposed algorithm, it was simulated compared to the existing method, and superior performance was confirmed compared to the existing method.

• ETRI Journal
• /
• v.32 no.5
• /
• pp.795-800
• /
• 2010

Embedding a large vocabulary speech recognition system in mobile devices requires a reduced acoustic model obtained by eliminating redundant model parameters. In conventional optimization methods based on the minimum description length (MDL) criterion, a binary Gaussian tree is built at each state of a hidden Markov model by iteratively finding and merging similar mixture components. An optimal subset of the tree nodes is then selected to generate a downsized acoustic model. To obtain a better binary Gaussian tree by improving the process of finding the most similar Gaussian components, this paper proposes a new distance measure that exploits the difference in likelihood values for cases before and after two components are combined. The mixture weight of Gaussian components is also introduced in the component merging step. Experimental results show that the proposed method outperforms MDL-based optimization using either a Kullback-Leibler (KL) divergence or weighted KL divergence measure. The proposed method could also reduce the acoustic model size by 50% with less than a 1.5% increase in error rate compared to a baseline system.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2012.05a
• /
• pp.623-625
• /
• 2012

The restoration of an image corrupted by Gaussian noise is an important task in image processing. There are many kinds of filters are proposed to remove Gaussian noise such as Gaussian filter, mean filter, weighted filter, etc. However, they perform not good enough for denoising and edge preservation. Hence, in this paper we proposed an adaptive weighted filter which considers spatial distance and the estimated variance of noise. We also compared the proposed method with existing methods through the simulation and used MSE(mean squared error) as the standard of judgement of improvement effect.