• 정보저장시스템학회논문집
• /
• 제6권1호
• /
• pp.18-24
• /
• 2010

The performance of the speech recognition mainly depends on uncertain factors such as speaker's conditions and environmental effects. The present study deals with the speech recognition of a number of multi-syllable isolated Korean words using soft computing techniques such as back-propagation neural network, fuzzy inference system, and fuzzy neural network. Feature patterns for the speech recognition are analyzed with 12th order thirty frames that are normalized by the linear predictive coding and Cepstrums. Using four models of speech recognizer, actual experiments for both single-speakers and multiple-speakers are conducted. Through this study, the recognizers of combined fuzzy logic and back-propagation neural network and fuzzy neural network show the better performance in identifying the speech recognition.

• 한국컴퓨터정보학회논문지
• /
• 제20권6호
• /
• pp.151-159
• /
• 2015

In traditional programing education, computers were used as the main tool. Consequently, it was problematic to provide education in an environment without computers or for learners without computer skills. To address this problem, this study developed and validated an unplugged robot education system capable of computerless programming education. The key feature of the proposed system is that programing can be done only by connecting programming blocks in symbols of a flow chart with built-in commands. Validation of the system was performed by a specialist group. Validity was very high with values of content validity ratio (CVR) over 0.7 in all evaluation criteria except "Ease of error debugging" and "Linkages to educational curriculum," whose CVR values were each 0.6. Future directions include improvement in the two areas that scored lower than the others did by, respectively, system improvement to support debugging in error conditions that may occur during the programming process, and development of user guide to support linkages to educational curriculum.

• Journal of information and communication convergence engineering
• /
• 제5권3호
• /
• pp.194-199
• /
• 2007

In this paper, a slotted CDMA_ALOHA protocol with hybrid ARQ is proposed for the wireless CDMA communication networks. The proposed protocol combines the characteristics of the slotted ALOHA, CDMA, and the hybrid ARQ, in order to increase the throughput by reducing the number of retransmissions when the channel experiences heavy traffic. The main feature of the proposed protocol is the utilization of the forward error correction capability to correct errors that appear after the CDMA dispreading of the packets. The base station does not need to ask so often for retransmission of erroneous packets. It will request for retransmission only when the FEC capability is exceeded. The performance of the proposed protocol is analyzed by considering the packet collision probability as well as the bit error probability. The numerical results show that the system throughput is closely related to the bit error rate of the wireless link and the FEC coding rate.

• 한국음향학회지
• /
• 제21권4호
• /
• pp.156-156
• /
• 2002

We apply independent component analysis (ICA) for extracting an optimal basis to the problem of finding efficient features for representing speech signals of a given speaker The speech segments are assumed to be generated by a linear combination of the basis functions, thus the distribution of speech segments of a speaker is modeled by adapting the basis functions so that each source component is statistically independent. The learned basis functions are oriented and localized in both space and frequency, bearing a resemblance to Gabor wavelets. These features are speaker dependent characteristics and to assess their efficiency we performed speaker identification experiments and compared our results with the conventional Fourier-basis. Our results show that the proposed method is more efficient than the conventional Fourier-based features in that they can obtain a higher speaker identification rate.

• 반도체디스플레이기술학회지
• /
• 제8권1호
• /
• pp.19-25
• /
• 2009

Abstract JPEG2000 achieve quality scalability through the rate control method used in the encoding process, which embeds quality layers to the code-stream. This architecture might raise two drawbacks. First, when the coding process finishes, the number and bit-rates of quality layers are fixed, causing a lack of quality scalability to code-stream encoded with a single or few quality layers. Second, in Post compression rate distortion (PCRD) the bit streams after the truncation points discarded. Therefore, computational power for the discarded bit streams is wasted. For solving of problem, through bit rate control, there are many researches. Each proposed algorithms have specially target feature that is improved performance like reducing computational power. Research results have strength and weakness. For the mean time, research contents are reviewed and compared, so we proposed research direction in the future.

• 한국정보처리학회논문지
• /
• 제7권3호
• /
• pp.1013-1020
• /
• 2000

In this paper, we propose an algorithm which segments hierarchically blocks of image using discrete cosine transform(DCT) and execute finite-state vector quantization (FSVQ) for each block. Using DCT coefficient feature, image is segmented hierarchically to large smooth block and small edge block, then the block hierarchy informations are transmitted. The codebooks are respectively constructed for each hierarchical blocks, the encoder transmits codeword index using FSVQ for reducing encoded bit with hierarchical segmentation. Compared with side match VQ(SMVQ) and hierarchical FSVQ(HFSVQ) algorithm, about Zelda and Boat image, the new algorithm shows better picture quality with 1.97dB and 2.85 dB difference as to SMVQ, 1.78dB and 1.85dB diffences as to HFSVQ respectively.

• 한국정보처리학회논문지
• /
• 제6권10호
• /
• pp.2816-2822
• /
• 1999

In this paper, a fast thinning method for fingerprint image by separating end and bifurcation region is proposed. To detect feature points in automatic fingerprint identification system, thinning of fingerprint is essential. The end and bifurcation regions in ridge line are separated by means of run-length coding, and parallel thinning method is applied to the separated regions. The rest parts except the end and bifurcation regions are processed by connecting center points of each run. The performance of the proposed method has been evaluated by CPU processing time and thinness measurement. By the experimental results, the proposed method is fast and has high thinness value.

• Journal of Information Processing Systems
• /
• 제19권1호
• /
• pp.98-108
• /
• 2023

This paper is an attempt to design segmentation method based on fully convolutional networks (FCN) and attention mechanism. The first five layers of the Visual Geometry Group (VGG) 16 network serve as the coding part in the semantic segmentation network structure with the convolutional layer used to replace pooling to reduce loss of image feature extraction information. The up-sampling and deconvolution unit of the FCN is then used as the decoding part in the semantic segmentation network. In the deconvolution process, the skip structure is used to fuse different levels of information and the attention mechanism is incorporated to reduce accuracy loss. Finally, the segmentation results are obtained through pixel layer classification. The results show that our method outperforms the comparison methods in mean pixel accuracy (MPA) and mean intersection over union (MIOU).

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2020년도 추계학술대회
• /
• pp.183-186
• /
• 2020

방대한 비디오 데이터의 지능형 분석을 수행하는 기계를 위한 비디오 부호화 기술의 필요성이 대두되면서 MPEG 에서는 VCM(Video Coding for Machine) 표준화를 시작하였다. VCM 은 지능형 머신(machine)의 임무 수행을 위한 비디오 또는 비디오 특징(feature)의 압축 표준 기술로 기술 탐색 단계의 표준화를 진행하고 있다. 본 논문에서는 머신비전(machine vision) 네트워크에서 추출되는 대용량의 특징 압축을 위한 전처리 단계로 보다 효과적인 특징 표현 방법을 제시한다. 제안하는 특징 표현 방법은 정규화, 양자화 과정을 거쳐 특징 데이터 크기를 감소시킨다. 실험에서 특징을 4 개의 값으로 양자화 했을 때, 원본 대비 16 배의 데이터 크기가 감소되지만 mAP 평가 성능은 35.4592 로 높은 수준으로 유지함을 확인하였다.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 춘계학술발표대회
• /
• pp.397-400
• /
• 2021

To solve the problem that existing computing methods cannot adequately represent the semantic features of sentences, Siamese TRAT, a semantic feature extraction model based on Transformer encoder is proposed. The transformer model is used to fully extract the semantic information within sentences and carry out deep semantic coding for sentences. In addition, the interactive attention mechanism is introduced to extract the similar features of the association between two sentences, which makes the model better at capturing the important semantic information inside the sentence. As a result, it improves the semantic understanding and generalization ability of the model. The experimental results show that the proposed model can improve the accuracy significantly for the semantic similarity calculation task of English and Chinese, and is more effective than the existing methods.