• Journal of information and communication convergence engineering
• /
• v.9 no.5
• /
• pp.487-490
• /
• 2011

This paper presents splitting and merging algorithm based on adaptive thresholding. The algorithm first divides the image into blocks, and then compares each block using the calculated thresholding value. The blocks which are same are merged using the certain threshold value and different blocks are split unless it satisfies the threshold value. When the block has been merged, maximum and minimum block sizes are determined then the average block size is determined. After the average block size is determined the average intensity and standard deviation of average block is calculated. The process of thresholding is applied to binarize the image. Finally, the experimental results show that the proposed method distinguishes clearly the background with text in the document image.

• English Language & Literature Teaching
• /
• v.16 no.3
• /
• pp.1-18
• /
• 2010

This study examined the effects of previewing and providing background knowledge on low level EFL readers' comprehension of expository texts and their responses to these treatments. 130 college freshmen were randomly placed into one of three treatment groups and read two expository texts reflecting unfamiliar cultural information. Prior to reading, one group was given previewing instruction, which included vocabulary preteaching and summaries, and a second group was provided with culture-specific background knowledge through watching videos and slides. The third group read each text without any prereading instruction. Immediately after reading a passage, subjects answered a 10-item multiple-choice test. Results showed significant positive effects of the previewing treatment and weak positive effects of the providing background knowledge treatment. Students' responses on the questionnaires revealed that the majority felt that the experimental treatments contributed to comprehension enhancement, made reading more enjoyable, and expedited their reading process. Students in the control group, however, indicated that they needed explicit prereading instruction in order to understand the texts. Pedagogical implications of the findings for EFL reading instruction are provided.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.885-886
• /
• 2006

The purpose of this study is to develop the optimal system(AAC device), which helps a person with a profound communication disorder to communicate with other people. Therefore, this system includes the user interface enhancement that is the user adaptation mode algorithm. The symbol is made with a text and an icon which is converted into Korean. The message contiol operates scanning and adjusts rate control of row-column scanning and linear scanning. This embedded system includes voice input/output and voice recording as well suggested method that could apply optimal device access algorithm from clinical environment. Therefore, we are experting that even the current system itself will be able to improve the life quality of people who need to communicate with the help of devices.

• Speech Sciences
• /
• v.12 no.4
• /
• pp.31-42
• /
• 2005

Recently, Gaussian Mixture Model (GMM), a special form of CHMM, has been applied to speaker identification and it has proved that performance of GMM is better than CHMM. Therefore, in this paper the speaker models based on GMM and a new GMM using the modified EM algorithm are introduced and evaluated for text-independent speaker identification. Various experiments were performed to evaluate identification performance of two algorithms. As a result of the experiments, the GMM speaker model attained 94.6% identification accuracy using 40 seconds of training data and 32 mixtures and 97.8% accuracy using 80 seconds of training data and 64 mixtures. On the other hand, the new GMM speaker model achieved 95.0% identification accuracy using 40 seconds of training data and 32 mixtures and 98.2% accuracy using 80 seconds of training data and 64 mixtures. It shows that the new GMM speaker identification performance is better than the GMM speaker identification performance.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.155-156
• /
• 2019

Traditional machine learning based emotion analysis methods such as Classification and Regression Tree (CART), Support Vector Machine (SVM), and k-nearest neighbor classification (kNN) are less accurate. In this paper, we propose an improved kNN classification method. Improved methods and data normalization achieve the goal of improving accuracy. Then, three classification algorithms and an improved algorithm were compared based on experimental data.

• Journal of Advanced Information Technology and Convergence
• /
• v.10 no.1
• /
• pp.45-56
• /
• 2020

Despite extensive research, performance enhancement of keyphrase (KP) extraction remains a challenging problem in modern informatics. Recently, deep learning-based supervised approaches have exhibited state-of-the-art accuracies with respect to this problem, and several of the previously proposed methods utilize Bidirectional Encoder Representations from Transformers (BERT)-based language models. However, few studies have investigated the effective application of BERT-based fine-tuning techniques to the problem of KP extraction. In this paper, we consider the aforementioned problem in the context of scientific articles by investigating the fine-tuning characteristics of two distinct BERT models - BERT (i.e., base BERT model by Google) and SciBERT (i.e., a BERT model trained on scientific text). Three different datasets (WWW, KDD, and Inspec) comprising data obtained from the computer science domain are used to compare the results obtained by fine-tuning BERT and SciBERT in terms of KP extraction.

• Asia Marketing Journal
• /
• v.21 no.3
• /
• pp.65-82
• /
• 2019

This study examines the relationship between social media advertisement and customer interest in the context of luxury brands. Further, this study investigates the effective ways to utilize visual types (pictorial advertisement and video advertisement) and contents types (website link and hash-tag) in social media advertising by proposing a time-series model to estimate the long-term effect of social media advertising on social search. We find that the pictorial advertisements are more effective than video advertisements, which provides a different result from previous existing research. In addition, advertisements using hashtags are more effective than web links due to efficiency of the search feature. Finally, since the number of brand fans also have a positive effect on advertising interest, it is essential to utilize social media advertising for the enhancement of customers' interests. Confirming that the effectiveness of social media advertising varies depending on how the visual contents and text are presented, this research can help marketing managers to assess predicted outcomes of using various methods of social media advertising.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.465-468
• /
• 2020

인터넷 미디어, OTT, VOD 등 신규미디어가 비장애인의 정보제공 매체로 널리 확대되나, 자막 서비스를 제공하지 않아 청각장애인의 정보 격차가 더욱 심화되고 있다. 청각장애인의 미디어 접근성 제고를 위해 음성인식 서버 및 스마트 폰·태블릿 앱 간 연계를 통해 음성을 인식하여 자동으로 자막을 생성하고 표시하는 음성-자막 자동 변환 시스템을 개발하였고 음성인식률을 높이기 위해 뉴스/시사/다큐 장르 영상 콘텐츠의 음성에 대해 학습용 데이터를 제작하여 음성인식 성능을 고도화 시켰다. 본 논문에서는 청각장애인을 위한 음성-자막 자동 변환시스템 구성과 음성인식률 비교 평가 결과를 보여준다.

• The KIPS Transactions:PartB
• /
• v.10B no.6
• /
• pp.673-680
• /
• 2003

In this paper, we study the enhancement of VQ (Vector Quantization) design for text independent speaker recognition. In a concrete way, we present a non-iterative method which makes a vector quantization codebook and this method performs non-iterative learning so that the computational complexity is epochally reduced The proposed Classified Space VQ (CSVQ) design method for text Independent speaker recognition is generalized from Semi-noniterative VQ design method for text dependent speaker recognition. CSVQ contrasts with the existing desiEn method which uses the iterative learninE algorithm for every traininE speaker. The characteristics of a CSVQ design is as follows. First, the proposed method performs the non-iterative learning by using a Classified Space Codebook. Second, a quantization region of each speaker is equivalent for the quantization region of a Classified Space Codebook. And the quantization point of each speaker is the optimal point for the statistical distribution of each speaker in a quantization region of a Classified Space Codebook. Third, Classified Space Codebook (CSC) is constructed through Sample Vector Formation Method (CSVQ1, 2) and Hyper-Lattice Formation Method (CSVQ 3). In the numerical experiment, we use the 12th met-cepstrum feature vectors of 10 speakers and compare it with the existing method, changing the codebook size from 16 to 128 for each Classified Space Codebook. The recognition rate of the proposed method is 100% for CSVQ1, 2. It is equal to the recognition rate of the existing method. Therefore the proposed CSVQ design method is, reducing computational complexity and maintaining the recognition rate, new alternative proposal and CSVQ with CSC can be applied to a general purpose recognition.

• Journal of the Korea Convergence Society
• /
• v.9 no.11
• /
• pp.61-68
• /
• 2018

In this paper, we propose a method of character detection by analyzing image enhancement and character type to detect characters in natural images that can be acquired in everyday life. The proposed method emphasizes the boundaries of the object part using the unsharp mask in order to improve the detection rate of the area to be recognized as a character in a natural image. By using the boundary of the enhanced object, the character candidate region of the image is detected using Maximal Stable Extermal Regions (MSER). In order to detect the region to be judged as a real character in the detected character candidate region, the shape of each region is analyzed and the non-character region other than the region having the character characteristic is removed to increase the detection rate of the actual character region. In order to compare the objective test of this paper, we compare the detection rate and the accuracy of the character region with the existing methods. Experimental results show that the proposed method improves the detection rate and accuracy of the character region over the existing character detection method.