• Journal of Korea Multimedia Society
• /
• v.8 no.10
• /
• pp.1347-1359
• /
• 2005

We propose an efficient block classification of the document images using the second-order statistical texture features computed from spatial gray level dependence matrix (SGLDM). We studied on the techniques that will improve the block speed of the segmentation and feature extraction speed and the accuracy of the detailed classification. In order to speedup the block segmentation, we binarize the gray level image and then segmented by applying smoothing method instead of using texture features of gray level images. We extracted seven texture features from the SGLDM of the gray image blocks and we applied these normalized features to the BP (backpropagation) neural network, and classified the segmented blocks into the six detailed block categories of small font, medium font, large font, graphic, table, and photo blocks. Unlike the conventional texture classification of the gray level image in aerial terrain photos, we improve the classification speed by a single application of the texture discrimination mask, the size of which Is the same as that of each block already segmented in obtaining the SGLDM.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.272-273
• /
• 2019

VVC(Versatile Video Coding)는 YUV 입력 영상에 대하여 Luma 성분과 Chroma 성분에 대하여 각각 다른 최적의 방법으로 블록분할 후 해당 블록에 대해서 화면 내 예측 또는 화면 간 예측을 수행하고, 예측영상과 원본영상의 차이를 변환, 양자화하여 압축한다. 이 과정에서 복원영상에는 블록화 노이즈, 링잉 노이즈, 블러링 노이즈 발생한다. 본 논문에서는 인코더에서 원본영상과 복원영상의 잔차신호에 대한 MAE(Mean Absolute Error)를 추가정보로 전송하여 이 추가정보와 복원영상을 이용하여 Deep Learning 기반의 신경망 네트워크로 영상의 품질을 높이는 방법을 제안한다. 복원영상의 노이즈를 감소시키기 위하여 영상을 $32{\times}32$블록의 임의로 분할하고, DenseNet기반의 UNet 구조로 네트워크를 구성하였다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.4
• /
• pp.775-782
• /
• 2006

We generated blocks from the direction-extracted fingerprint during the pre-process of the fingerprint recognition algorithm and performed training by using the direction minutiae of each block as the input pattern of the neural network, so that we extracted the core points to use in the matching. Based on this, we designed the fingerprint recognition embedded system and tested it using the control board and the serial communication to utilize it for a variety of application systems. As a result, we can verify the reliance satisfactorily.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2006.11a
• /
• pp.145-148
• /
• 2006

본 논문은 $0.35{\mu}m$ 2중 폴리 CMOS 공정을 이용하여 프리만 신경회로 모델의 기본 요소가 되는 입력 취합 블록과 필드 앤드 홀드 방식의 2차 저역 통과 필터의 구현 및 부궤환과 비대칭 트랜스 콘덕터로 이루어지는 비선형 함수 블록을 설계하고 SPICE 회로 모의실험을 통해 결과를 확인하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.171-173
• /
• 2016

128-비트의 마스터 키를 지원하는 블록암호 AES-128을 IoT 보안에 적합하도록 경량화하여 구현하였다. 키 스케줄러와 라운드 블록을 8 비트 데이터 패스로 구현하고, 다양한 최적화 방법을 적용함으로써 하드웨어를 최소화시켰으며, 100 MHz 클록 주파수에서 4,400 GE의 작은 게이트로 구현되었다. Verilog HDL로 설계된 AES 크립토 코어를 Vertex5 XC5VSX50T FPGA 디바이스에 구현하여 올바로 동작함을 확인하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.9A
• /
• pp.1411-1418
• /
• 1999

In this paper, we propose a fast motion estimation algorithm using motion prediction and neural network. Considering that the motion vectors have high spatial correlation, the motion vector of current block is predicted by those of neighboring blocks. The codebook of motion vector is designed by Kohonen self-organizing feature map(KSFM) learning algorithm which has a fast learning speed and 2-D adaptive chararteristics. Since the similar codevectors are closely located in the 2-D codebook the motion is progressively estimated from the predicted codevector in the codebook. Computer simulation results show that the proposed method has a good performance with reduced computational complexity.

• Journal of KIISE:Software and Applications
• /
• v.29 no.7
• /
• pp.425-435
• /
• 2002

This paper introduces a macroblock-based rate control algorithm using the neural classifier based in Self Organization feature Maps (SOFM). In contrast to the conventional rate control methods based on the mathematical rate distortion (RD) model and the feedback regression, proposed method can actively adapt to the rapid-varying image characteristics by establishing the global model for bitrate control and by using the SOFM based neural classifier to manage that model. Proposed rate control algorithm has 0.2 dB ~ 0.6 dB better performances than MPEG-4 macroblock-based rate control algorithm by evaluating with the encoded Peak Signal to Noise Ratios while maintaining similar overall computational complexity.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.269-271
• /
• 2022

In recent years, RNN networks with LSTM blocks have been used extensively in machine learning tasks that process sequential data. These networks have proven to be particularly good at sequential language processing tasks by being more able to accurately predict the next most likely word in a given sequence than traditional neural networks. This study trained an RNN / LSTM neural network on three different translations of 150 biblical Psalms - in both English and Korean. The resulting model is then fed an input word and a length number from which it automatically generates a new Psalm of the desired length based on the patterns it recognized while training. The results of training the network on both English text and Korean text are compared and discussed.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.4
• /
• pp.234-240
• /
• 2022

This paper proposes a customized AI exercise recommendation service for balancing the relative amount of exercise according to the working environment by each occupation. WISDM database is collected by using acceleration and gyro sensors, and is a dataset that classifies physical activities into 18 categories. Our system recommends a adaptive exercise using the analyzed activity type after classifying 18 physical activities into 3 physical activities types such as whole body, upper body and lower body. 1 Dimensional convolutional neural network is used for classifying a physical activity in this paper. Proposed model is composed of a convolution blocks in which 1D convolution layers with a various sized kernel are connected in parallel. Convolution blocks can extract a detailed local features of input pattern effectively that can be extracted from deep neural network models, as applying multi 1D convolution layers to input pattern. To evaluate performance of the proposed neural network model, as a result of comparing the previous recurrent neural network, our method showed a remarkable 98.4% accuracy.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.05a
• /
• pp.404-406
• /
• 2015

This paper describes a small-area hardware implementation of the block cipher algorithm CLEFIA-128 which supports for 128-bit master key. A compact structure using single data processing block is adopted, which shares hardware resources for round transformation and the generation of intermediate values for round key scheduling. In addition, data processing and key scheduling blocks are simplified by utilizing a modified GFN(generalized Feistel network) and key scheduling scheme. The CLEFIA-128 crypto-processor is verified by FPGA implementation. It consumes 823 slices of Virtex5 XC5VSX50T device and the estimated throughput is about 105 Mbps with 145 MHz clock frequency.