• 인지과학
• /
• 제21권1호
• /
• pp.145-156
• /
• 2010

본 논문은 원본 디지털 영상을 보호하고 동시에 저작권 정보 등을 저장할 수 있는 콘텐츠 기반 정지 영상 보호에 관한 새로운 방법을 제시한다. Gradient값을 이용한 기존의 알고리즘은 픽셀 단위의 소벨 연산자 알고리즘이 적용되었기 때문에 외부 공격에 대하여 상대적으로 민감하게 반응하고 저작권 정보 등을 정확하게 검출하지 못하는 단점이 있다. 따라서 본 논문에서는 이러한 약점을 보완하기 위하여 블록 단위의 연산이 이루어지는 OCL(Orientation Certainty Level)을 적용하여 특징점을 선택하였다. 실험 결과 기존의 알고리즘에 비해 변화가 심한 공격에서도 99% 이상의 높은 검출도를 나타내는 것을 볼 수 있고 특히 회전 공격에 대해서는 10%이상의 큰 성능 향상을 보여주고 있다.

• 대한원격탐사학회지
• /
• 제39권5_3호
• /
• pp.891-907
• /
• 2023

Compact Advanced Satellite 500-4 (CAS500-4) is scheduled to be launched to collect high spatial resolution data focusing on vegetation applications. To achieve this goal, accurate surface reflectance retrieval through atmospheric correction is crucial. Therefore, a machine learning-based atmospheric correction algorithm was developed to simulate atmospheric correction from a radiative transfer model using Sentinel-2 data that have similarspectral characteristics as CAS500-4. The algorithm was then evaluated mainly for forest areas. Utilizing the atmospheric correction parameters extracted from Sentinel-2 and GEOKOMPSAT-2A (GK-2A), the atmospheric correction algorithm was developed based on Random Forest and Light Gradient Boosting Machine (LGBM). Between the two machine learning techniques, LGBM performed better when considering both accuracy and efficiency. Except for one station, the results had a correlation coefficient of more than 0.91 and well-reflected temporal variations of the Normalized Difference Vegetation Index (i.e., vegetation phenology). GK-2A provides Aerosol Optical Depth (AOD) and water vapor, which are essential parameters for atmospheric correction, but additional processing should be required in the future to mitigate the problem caused by their many missing values. This study provided the basis for the atmospheric correction of CAS500-4 by developing a machine learning-based atmospheric correction simulation algorithm.

• 한국컴퓨터정보학회논문지
• /
• 제26권1호
• /
• pp.1-9
• /
• 2021

OpenGL compute shader는 다른 shader 단계와 다르게 동작하며, 병렬로 모든 데이터를 계산하는데 사용할 수 있다. 본 논문은 OpenGL compute shader에서 반복 켤레 기울기 방법을 통해 희소선형 시스템을 계산하기 위한 GPU 기반의 병렬 알고리즘 제안하였다. 제안된 희소 선형 해결 방법은 대칭인 양의 정부호 행렬과 같은 대형 선형 시스템을 해결하기 위해 사용된다. 본 논문은 이 알고리즘을 사용하여 매트릭스 형식이 다른 8가지 예제들에 대해서 CPU와 GPU를 기반으로한 성능 비교 결과를 제공한다. 본 논문은 4가지 잘 알려져 있는 매트릭스 형식(Dense, COO, ELL and CSR)을 매트릭스 저장소를 사용하였다. 8개의 희소 매트릭스를 사용한 성능 비교 실험에서 GPU 기반 선형 해결 시스템이 CPU 기반 선형 해결 시스템보다 훨씬 빠르며, GPU 기반에서 0.64ms, CPU 기반에서 15.37ms의 평균 컴퓨팅 시간을 제공한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권9호
• /
• pp.3745-3761
• /
• 2020

Low-rank representation methods already achieve many applications in the image reconstruction. However, for high-gradient image patches with rich texture details and strong edge information, it is difficult to find sufficient similar patches. Existing low-rank representation methods usually destroy image critical details and fail to preserve edge structure. In order to promote the performance, a new representation-based image super-resolution reconstruction method is proposed, which combines gradient domain guided image filter with the structure-constrained low-rank representation so as to enhance image details as well as reveal the intrinsic structure of an input image. Firstly, we extract the gradient domain guided filter of each atom in high resolution dictionary in order to acquire high-frequency prior information. Secondly, this prior information is taken as a structure constraint and introduced into the low-rank representation framework to develop a new model so as to maintain the edges of reconstructed image. Thirdly, the approximate optimal solution of the model is solved through alternating direction method of multipliers. After that, experiments are performed and results show that the proposed algorithm has higher performances than conventional state-of-the-art algorithms in both quantitative and qualitative aspects.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2006년도 하계종합학술대회
• /
• pp.817-818
• /
• 2006

In Magnetic Resonance Imaging(MRI), the QRS complex of ECG is used as a trigger signal for MRI scan. But, gradient and RF(radio frequency) artifacts which are caused to static and dynamic field in MRI scanner cause interference in the ECG. Also, the signal shape of theses artifacts can be similar to the QRS-complex, causing possible misinterpretation during patient monitoring and false gating of the MRI. In case of using general FIR or IIR band-pass filters for minimizing the artifacts, artifact-reduction-ratio is not excellent. So, an adaptive real-time digital filter is proposed for reduction of noise by gradient and RF(radio frequency) artifacts. The proposed filter for MRI-Gating is based on the noise-canceller with NLMS(Normalized Least Mean Square) algorithm. The reference signals of the adaptive noise canceller are a combination of the noisy three channel ECG signals. In conclusions, the proposed method showed the acceptable quality of ECG signal with sufficient SNR for gating the MRI and possibility of real time implementation.

• 한국산학기술학회논문지
• /
• 제15권5호
• /
• pp.3132-3138
• /
• 2014

많은 통신 시스템에서 심볼간 간섭, 직류성 및 충격성 잡음은 해결하기 어려운 문제로 남아있다. 이러한 간섭신호들을 제거하기 위하여 확률분포 상관도 개념이 블라인드 Equalization에 사용되었다. 그러나 이 알고리듬은 과다한 계산량이 문제로 남아 있다. 이 논문에서는 확률분포 상관도에 기반한 블라인드 알고리듬의 반복적 계산 방법을 제안하였다. 비용함수의 기울기 계산에 쓰이는 합산 계산을 반복적 방식으로 기울기를 계산하도록 바꾸었다. 이 방식은 M 개의 송신 심볼에 대해 N 개의 블록 샘플들을 가지고 계산하는 기존 알고리듬의 계산량 O(NM)으로부터 O(M)으로 그 계산량을 획기적으로 줄인다. 따라서 현실적 구현의 장점을 가지면서 동시에 잡음 및 간섭에 대한 강인성을 그대로 유지한다. 시뮬레이션 결과에서도 제안한 방식은 줄여진 계산량으로 동일한 학습 성능을 보였다.

• 한국음향학회지
• /
• 제32권3호
• /
• pp.279-285
• /
• 2013

본 논문에서는 기존의 global soft decision 기법에서 음성 부재 확률을 구할 때의 음성 부재와 존재에 대한 a priori 확률값의 비(q)에 스펙트럼 변이 기법을 적용한 음성 향상 기법을 제안한다. 기존의 global soft decision 방법은 음성 부재 확률을 구하기 위해 가정한 가설에 따라 고정된 q 값을 사용하였지만, 본 논문에서 제안한 알고리즘은 기존의 고정된 값에 직전 2 프레임에서의 음성 존재 여부와 스펙트럼 변이 값의 상태 조건에 따라 적응적으로 q 값이 가변되도록 하여 음성 부재 확률을 향상시키는 기법이다. 제안된 방법의 성능 평가를 위해 ITU-T P.862 PESQ(Perceptual Evaluation of Speech Quality)를 이용하여 평가하였고, 그 결과 제안된 스펙트럼 변이 기법을 적용한 방법이 기존의 global soft decision 방법보다 향상된 결과를 보여주었다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.663-667
• /
• 2004

This paper presents new pruning and learning methods for the fuzzy rule-based classifier. The structure of the proposed classifier is framed from the fuzzy sets in the premise part of the rule and the Bayesian classifier in the consequent part. For the simplicity of the model structure, the unnecessary features for each fuzzy rule are eliminated through the iterative pruning algorithm. The quality of the feature is measured by the proposed correctness method, which is defined as the ratio of the fuzzy values for a set of the feature values on the decision region to one for all feature values. For the improvement of the classification performance, the parameters of the proposed classifier are finely adjusted by using the gradient descent method so that the misclassified feature vectors are correctly re-categorized. The cost function is determined as the squared-error between the classifier output for the correct class and the sum of the maximum output for the rest and a positive scalar. Then, the learning rules are derived from forming the gradient. Finally, the fuzzy rule-based classifier is tested on two data sets and is found to demonstrate an excellent performance.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제10권11호
• /
• pp.5497-5512
• /
• 2016

An improved optimal weights based on parallel interference cancellation algorithm has been proposed to cancel for interference induced by multi-user access satellite to ground uplink system. Due to differences in elevation relative motion between the user and the satellite, as well as access between users, resulting in multi-user access interference (Multi-user Access Interference, MUI), which significantly degrade system performance when multi-user access. By steepest gradient method, it obtained based on the MMSE criterion, parallel interference cancellation adjust optimal weights to obtain the maximum SINR. Compared to traditional parallel interference cancellation (Parallel Interference Cancellation, PIC) algorithm or serial interference cancellation ( Successive interference Cancellation, SIC), the accuracy of which is not high and too many complex iterations, we establish the multi-user access to the satellite to ground up link system to demonstrate that the improved WPIC algorithm could be provided with high accuracy and relatively low number of iterations.

• 한국측량학회지
• /
• 제33권6호
• /
• pp.569-577
• /
• 2015

The navigation algorithm developed based on the extended Kalman filter (EKF) sometimes diverges when the linearity between the measurements and the states is not preserved. In this study, new update conditions together with two conditions from previous study for gravity gradient referenced navigation (GGRN) were deduced for the filter performance. Also, the effect of each update conditions was evaluated imposing the various magnitudes of the database (DB) and the sensor errors. In case the DB and the sensor errors were supposed to 0.1 Eo and 0.01 Eo, the navigation performance was improved in the eight trajectories by using part of gravity gradient components that independently estimate states located within trust boundary. When applying only the components showing larger variation, around 200% of improvement was found. Even the DB and sensor error were supposed to 3 Eo, six update conditions improved performance in at least seven trajectories. More than five trajectories generated better results with 5 Eo error of the DB and the sensor. Especially, two update conditions successfully control divergence, and bounded the navigation error to the 1/10 level. However, these update conditions could not be generalized for all trajectories so that it is recommended to apply update conditions at the stage of planning, or as an index of precision of GGRN when combine with various types of geophysical data and algorithm.