• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.1
• /
• pp.61-68
• /
• 2015

In recent times, the median filtering (MF) detector as a forensic tool for the recovery of forgery images' processing history has concerned broad interest. For the classification of MF image, MF detector should be designed with smaller feature set and higher detection ratio. This paper presents a novel method for the detection of MF in altered images. It is transformed from BMP to several kinds of MF image by the median window size. The difference distribution values are computed according to the window sizes and then the values construct the feature set same as the MF window size. For the MF detector, the feature set transformed to the model specification which is computed using latent growth modeling (LGM). Through experiments, the test image is classified by the discriminant into two classes: the true positive (TP) and the false negative (FN). It confirms that the proposed algorithm is to be outstanding performance when the minimum distance average is 0.119 in the confusion of TP and FN for the effectivity of classification.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.4
• /
• pp.497-501
• /
• 2010

A software montage means information that can be extracted quickly from software and includes inherent characteristics. If a montage is made from well-known programs, we can filter candidates of similar programs among the group of programs based on the montage. In this paper, we suggest software montages based on two characteristics: API calls and strings. To evaluate the proposed montages, we performed experiments to filter candidates of some similar programs to instant messenger programs. From the experiments, we confirmed that the proposed montages can be used as a forensic tool that filters a group of similar programs even when their functions are not known in advance.

• The Journal of the Acoustical Society of Korea
• /
• v.12 no.6
• /
• pp.13-20
• /
• 1993

충격 응답 시간이 긴 시스템을 모델링하기 위한 실시간 적응 디지털 필터를 구현하였다. 대상 시스템의 충격 응답 시간이 길 때, 일반적인 적응 디지털 필터를 사용하는 경우 발생하는 수렴 속도 저하와 계산량 증가 문제를 해결하기 위해서 서브밴드 구조를 갖는 적응 디지털 필터를 구성하였다. 실시간 처리 시스템에서는 GQMF을 사용하여 입력 신호를 4개 대역으로 분할하여 각 대역별로 적응 필터링을 수행함으로써 수렴 속도를 향상시킨다. 또한 대역별 신호를 동시에 분산 처리하기 때문에 계산량 면에서 효율적이므로 시스템의 충격 응답이 긴 경우에는 실시간 처리가 가능하다. 하드웨어 구성은 범용 신호 처리 프로세서인 DSP56001을 호스트 프로세서로 사용하며, 적응 디지털 필터 칩 DSP56200을 사용하여 각 대역 적응 필터를 구성하였다. 실험은 충격 응답 시간이 16 kHz 필터링 시 2000 탭 길이로 가정된 시스템을 대상으로 부동 소수점 시뮬레이션 결과와 실시간 처리 시스템의 결과를 비교하였다. 밴드를 나누지 않은 기존의 방법과 서브밴드 시스템의 비교 실험 결과 입력이 백색 잡음인 경우 대역별 간섭에 의한 성능 저하가 있었으나, 음성과 유사한 특성을 갖는 유색 잡음인 경우 서브밴드 시스템이 단일 시스템에 비해 성능 향상을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.109-114
• /
• 2022

질의 생성 모델은 스마트 스피커, 챗봇, QA 시스템, 기계 독해 등 다양한 서비스에 사용되고 있다. 모델을 다양한 서비스에 잘 적용하기 위해서는 사용자들의 실제 질의 특성을 반영한 자연스러운 질의를 만드는 것이 중요하다. 본 논문에서는 사용자 질의 특성을 반영한 간결하고 자연스러운 질의 자동 생성 모델을 소개한다. 제안 모델은 topic 키워드를 통해 모델에게 생성 자유도를 주었으며, 키워드형 질의→자연어 질의→응답으로 연결되는 chain-of-thought 형태의 다중 출력 구조를 통해 인과관계를 고려한 결과를 만들도록 했다. 최종적으로 MRC 필터링과 일관성 필터링을 통해 고품질 질의를 선별했다. 베이스라인 모델과 비교해 제안 모델은 질의의 유효성을 크게 높일 수 있었다.

• Journal of KIISE:Databases
• /
• v.29 no.6
• /
• pp.453-463
• /
• 2002

With the prevalence of multi-dimensional data such as images, content-based retrieval of data is becoming increasingly important. To handle multi-dimensional data, multi-dimensional index structures such as the R-tree, Rr-tree, TV-tree, and MVP-tree have been proposed. Numerous research results on how to effectively manipulate these structures have been presented during the last decade. Query processing strategies, which is important for reducing the processing time, is one such area of research. In this paper, we propose query processing algorithms for R-tree based structures. The novel aspect of these algorithms is that they make use of the notion of VP filtering, a concept borrowed from the MVP-tree. The filtering notion allows for delaying of computational overhead until absolutely necessary. By so doing, we attain considerable performance benefits while paying insignificant overhead during the construction of the index structure. We implemented our algorithms and carried out experiments to demonstrate the capability and usefulness of our method. Both for range query and incremental query, for all dimensional index trees, the response time using VP filtering was always shorter than without VP filtering. We quantitatively showed that VP filtering is closely related with the response time of the query.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.2
• /
• pp.253-258
• /
• 2012

A computationally efficient interpolation filter with a low-distortion performance is a key component to utilize the naturally-sampled pulse width modulation (NPWM) in a digital domain. To realize the efficient interpolation filter, we propose a novel design based on the recently-proposed modified Farrow filter. The proposed filter shows a better pass-band distortion performance maintaining similar degree of complexity compared with the conventional Lagrange interpolation filter. We achieve the maximum distortion deviation of 10-3 dB to 20-kHz audible frequency range and distortion reduction of 1/6 times compared with the Lagrange interpolation filter.

• Science of Emotion and Sensibility
• /
• v.25 no.1
• /
• pp.67-78
• /
• 2022

Interest in bio-signal monitoring of wearable devices is increasing significantly as the next generation needs to develop new devices to dominate the global market of the information and communication technology industry. Accordingly, this research developed a resistive textile strain sensor through a wetting process in a single-wall carbon nanotube dispersion solution using an E-Band with low hysteresis. To measure the resistance signal in the E-Band to which electrical conductivity is applied, a universal material tester, an Arduino, and LCR meters that are microcontroller units were used to measure the resistance change according to the tensile change. To effectively handle various noises generated due to the characteristics of the fabric textile strain sensor, the filter performance of the sensor was evaluated using the moving average filter, Savitsky-Golay filter, and intermediate filters of signal processing. As a result, the reliability of the filtering result of the moving average filter was at least 89.82% with a maximum of 97.87%, and moving average filtering was suitable as the noise filtering method of the textile strain sensor.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2014.06a
• /
• pp.107-108
• /
• 2014

부화소 기반 영상 축소기법은 각각의 부화소를 조절함으로써 픽셀 기반 영상 축소기법보다 해상도를 향상시킬 수 있는 방법이다. 그러나 부화소에 의한 해상도의 증가는 종종 색상정보의 왜곡을 발생시킨다. 부화소 랜더링의 주요과제는 선명도를 유지함과 동시에 색조왜곡현상을 억제하는 것이다. 선행연구들은 부화소랜더링을 위해 1 차원 혹은 2 차원 필터를 최적화 하였지만, 지역적인 특성을 고려하지 않았기 때문에 출력영상의 화질이 저하되는 현상이 발생한다. 본 논문은 위와 같은 문제를 해결하기 위해 내용적응형 2D FIR 필터를 제작방법을 제안한다. 제안필터는 충분한 수의 저해상도 패치와 고해상도 패치 쌍을 이용하여 임의의 고해상도 패치로부터 고화질의 저해상도 패치를 만들기 위한 최적의 내용적응형 2D FIR 필터를 학습한다. 학습된 필터에 의한 실험결과 제안하는 필터가 종례기법들 보다 색조왜곡현상이 현저히 줄어들고, 출력영상의 선명도를 유지함을 보여준다.

• Journal of the Korean Society of Radiology
• /
• v.2 no.4
• /
• pp.33-40
• /
• 2008

Images acquired by a digital X-ray imaging system are inherently degraded due to system degradation process and additive noise sources. The system degradation in image quality is typically described as the system response function characterized by the modulation transfer function (MTF) and the noise term described as the noise power spectrum (NPS). In this case, we can restore the blur image as close as possible to the original image by using modified filtering designed for digital imaging system, as we know more precisely about the MTF and the NPS. In this paper, by performing simulation, we tried to restore blurred images taken from a digital X-ray imaging system based upon conventional filtering techniques such as a direct-inverse filtering, limited-inverse filtering, or a Wiener filtering, and evaluated the characteristics of the image restoration.

• Journal of Korea Multimedia Society
• /
• v.9 no.2
• /
• pp.255-263
• /
• 2006

It is common that a junk mail has an inconsistency of mail addresses between those of the mail headers and the mail recipients. In addition, users easily know that an email is a junk or legitimate mail only by looking for the title of the email. In this paper, we tried to apply the filtering classifiers of mail address validation check and the combination method of title-content weighting to improve the performance of junk mail filtering system. In order to verify the effectiveness of the proposed method, we performed an experiment by applying them to Naive Bayesian classifier. The experiment includes the unit testing and the combination of the filtering techniques. As a result, we found that our method improved 11.6% of recall and 2.1% of precision that it contributed the enhancement of the junk mail filtering system.