• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.13 no.3
• /
• pp.221-230
• /
• 1988

This paper presents an effective block matching algorithm(BMA) in which the number of search point is about a half of that of three step search, and the number of search step is fixed a four. The performance of the proposed algorithm is compared with those of three step search and one-at-at time search(OTS) for three video sequences composed of 16 framse. Moreover the performance of applying subsampling or integral projection to each BMA for further reducing the amount of computation is considered.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2005.05a
• /
• pp.801-804
• /
• 2005

본 논문에서는 Discrete Wavelet Transform(DWT) 기반의 워터마크 알고리즘을 제안하였다. 제안된 방법은 원 이미지를 4 개의 subimages 로 나누고, DWT 후 한 개의 subimage 의 저주파 영역에 워터마크를 삽입하였다. Subsampling 방법을 사용해 원 이미지 없이 워터마크를 추출하였다. 워터마크는 저주파($LL_2$) 영역에 삽입해 외부의 공격에 강인한 성격을 가지도록 하였고, 화질 열화도 줄일 수 있었다. 잘 알려진 이미지에 대한 실험을 통해 본 논문의 알고리즘의 타당성을 입증하였다.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.1
• /
• pp.1-8
• /
• 2016

Convolutional neural network (CNN) consists of a few pairs of both convolution layer and subsampling layer. Thus it has more hidden layers than multi-layer perceptron. With the increased layers, the size of convolution mask ultimately determines the total number of weights in CNN because the mask is shared among input images. It also is an important learning factor which makes or breaks CNN's learning. Therefore, this paper proposes the best method to choose the convolution size and the number of layers for learning CNN successfully. Through our face recognition with vast learning examples, we found that the best size of convolution mask is 5 by 5 and 7 by 7, regardless of the number of layers. In addition, the CNN with two pairs of both convolution and subsampling layer is found to make the best performance as if the multi-layer perceptron having two hidden layers does.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.508-510
• /
• 2001

동영상을 효율적으로 압축하기 위한 움직임백터 예측에 관한 많은 연구가 진행되어 왔다. 가장 일반적인 FBMA(Full search-based Block Matching Algorithm)는 화질은 좋지만 계량이 많기 때문에 실시간 인코딩을 요구하는 시스템에서 사용하는데 문제가 있다. 좋은 화질을 유지하면서 인코딩 속도를 해결하기 위한 많은 알고리즘들이 제안되어 왔지만 ASIC이나 소형 시스템에서 사용할 수 있는 방법이 계속 요구되고 있다. 본 논문에서는 계산량을 더욱 줄여 속도향상을 위한 방법인 TSWS(Two-Step search With Subsampling method) 제안하였다. TSWS는 블록정합알고리즘에 기반을 두고 있으며, 서브샘플링한 값으로 움직임 벡터를 찾는다. TSWS를 사용하였을 때 기존 방법들이 제공하는 주관적 화질이나 PSNR을 어느 정도 유지하면서도 속도를 20-30% 정도 개선시킬 수 있다.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.44 no.12
• /
• pp.77-84
• /
• 2007

In this parer, a wideband low-noise amplifier (LNA) has been realized in a 0.18mm CMOS technology for the applications of subsampling direct-conversion RF receivers. By exploiting the inverter-type transimpedance input stage with a 3rd-order Chebyshev matching network, the wideband LNA demonstrates the measured results of the -3dB bandwidth of 5.35GHz, the power gain (S21) of $12\sim18dB$, the noise figure (NF) of $6.9\sim10.8dB$, and the broadband input/output impedance matching of less than -10dB/-24dB within the bandwidth, respectively. The chip dissipates 32.4mW from a single 1.8V supply, and occupies the area of $0.56\times1.0mm^2$.

• Journal of the Korean Statistical Society
• /
• v.32 no.3
• /
• pp.213-224
• /
• 2003

We give an overview of results which have appeared or will appear elsewhere demonstrating that by suitably modifying the bootstrap principle, its applicability can be greatly enhanced. Although we state our results for the iid case, extensions are, at least heuristically, easy.

• ETRI Journal
• /
• v.32 no.2
• /
• pp.214-221
• /
• 2010

Radio frequency (RF) subsampling can be used by radio receivers to directly down-convert and digitize RF signals. A goal of a cognitive radio/software defined ratio (CR/SDR) receiver design is to place the analog-to-digital converter (ADC) as near the antenna as possible. Based on this, a band-pass sampling (BPS) frontend for CR/SDR is proposed and verified. We present a receiver architecture based second-order BPS and signal processing techniques for a digital RF frontend. This paper is focused on the benefits of the second-order BPS architecture in spectrum sensing over a wide frequency band range and in multiband receiving without modification of the RF hardware. Methods to manipulate the spectra are described, and reconstruction filter designs are provided. On the basis of this concept, second-order BPS frontends for CR/SDR systems are designed and verified using a hardware platform.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.11 no.3
• /
• pp.163-171
• /
• 2016

Lane detection is a widely researched topic. Although simple road detection is easily achieved by previous methods, lane detection becomes very difficult in several complex cases involving noisy edges. To address this, we use a Convolution neural network (CNN) for image enhancement. CNN is a deep learning method that has been very successfully applied in object detection and recognition. In this paper, we introduce a robust lane detection method based on a CNN combined with random sample consensus (RANSAC) algorithm. Initially, we calculate edges in an image using a hat shaped kernel, then we detect lanes using the CNN combined with the RANSAC. In the training process of the CNN, input data consists of edge images and target data is images that have real white color lanes on an otherwise black background. The CNN structure consists of 8 layers with 3 convolutional layers, 2 subsampling layers and multi-layer perceptron (MLP) of 3 fully-connected layers. Convolutional and subsampling layers are hierarchically arranged to form a deep structure. Our proposed lane detection algorithm successfully eliminates noise lines and was found to perform better than other formal line detection algorithms such as RANSAC

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.45 no.11
• /
• pp.89-94
• /
• 2008

In this paper, a cascade of a wideband amplifier and a high-Q bandpass filter (BPF) has been realized in a 0.18mm CMOS technology for the applications of subsampling direct-conversion receivers. The wideband amplifier is designed to obtain the -3dB bandwidth of 5.4GHz, and the high-Q BPF is designed to select a 2.4GHz RF signal for the Bluetooth specifications. The measured results demonstrate 18.8dB power gain at 2.34GHz with 31MHz bandwidth, corresponding to the quality factor of 75. Also, it shows the noise figure (NF) of 8.6dB, and the broadband input matching (S11) of less than -12dB within the bandwidth. The whole chip dissipates 64.8mW from a single 1.8V supply and occupies the area of $1.0{\times}1.0mm2$.

• The Korean Journal of Applied Statistics
• /
• v.14 no.2
• /
• pp.429-437
• /
• 2001

단순 이단계 표본 추출의 경우에 최적 선택률은 Hansen과 Hurwitz(1949)에 의하여 구하여졌다. 그러나 통계청에서 실시하는 표본조사등은 층화 이단계 추출을 한다. 따라서 실제적인 필요성에 의하여 층화 2단계 표본 설계를 시도 하였다. 층화 이단계 표본추출시에 주어진 비용아래서 모총계의 추정량의 분산을 최소로 하는 최적의 선택확률(optimum selection probability), 표본추출율과 부차 표본추출율을 Lagrangean 승수법에 의하여 구한다.