• 한국멀티미디어학회논문지
• /
• 제21권8호
• /
• pp.849-857
• /
• 2018

In this paper, we propose an automatic composition method using time series embedding of RNN Auto-Encoder. RNN Auto-Encoder can learn existing songs and can compose new songs from the trained RNN decoder. If one song is fully trained in the RNN Auto-Encoder, the song is embedded into the vector values of RNN nodes in the Auto-Encoder. If we train a lot of songs and apply a specific vector to the decoder of Auto-Encoder, then we can obtain a new song that combines the features of trained multiple songs according to the given vector. From extensive experiments we could find that our method worked well and generated various songs by selecting of the composition vectors.

• 한국정보통신학회논문지
• /
• 제9권2호
• /
• pp.330-337
• /
• 2005

본 논문에서는 SVM(Support Vector Machine)을 이용하여 얼굴 검출 성능을 향상시키는 방법을 제안한다. 본 논문에서는 먼저 영상내의 거대한 특징 집합으로부터 중요한 작은 특징 집합을 선택하는 AdaBoost 기반 객체 검출 방법을 사용하여 얼굴 후보 영역을 검출한다. 그 다음에는 특징 벡터에 대해 SVM 기반 이진분류를 수행하여 후보 영역의 영상이 얼굴인지 아닌지를 판별한다 실험 결과 본문에서 제안한 방법은 기존의 방법에 비하여 얼굴 검출의 정확도를 향상시켰다.

• 전자공학회논문지A
• /
• 제32A권1호
• /
• pp.234-243
• /
• 1995

An arry circuit is designed for parallel computation of vector-radix 2-D discrete cosine transform (VR-FCT) which is a fast algorithm of DCT. By using a 2-D array of processing elements (PEs), the butterfly structure of the VR-FCT can be efficiently implemented with high condurrency and local communication geometry. The proposed implementation features architectural medularity, regularity and locality, so that it is very suitable for VLSI realization. Also, no transposition memory is required. The array core for (8$\times$8) 2-D DCT, which is designed usign ISRC 1.5.mu.m N-Well CMOS technology, consists of 64 PEs arranged in (8$\times$8) 2-D array and contains about 98,000 transistors on an area of 138mm$^{2}$. From simulation results, it is estimated that (8$\times$8) 2-D DCT can be computed in about 0.88 .mu.sec at 50 MHz clock frequency, resulting in the throughput rate of about 72${\times}10^[6}$ pixels per second.

• 제어로봇시스템학회논문지
• /
• 제6권5호
• /
• pp.390-395
• /
• 2000

We present a fingerprint identification algorithm using the wavelet transform and correlation. The wavelet transform is used because of its simple operation to extract fingerprint minutiaes features for fingerprint classification. We perform the rowwise 1-D wavelet transform for a $256\times256$ fingerprint image to get a $1\times256$ column vector using the Haar wavelet and repeat 1-D wavelet transform for a 1$\times$256 column vector to get a $1\times4$ feature vector. Using PNN(Probabilistic Neural Network), we select the possible candidates from the stored feature vectors for fingerprint images. For those candidates, we compute the correlation between the input binary image and the target binary image to find the most similar fingerprint image. The proposed algorithm may be the key to a low cost fingerprint identification system that can be operated on a small computer because it does not need a large memory size and much computation.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2009년도 IWAIT
• /
• pp.435-440
• /
• 2009

The scalable vector graphics (SVG) standard has allowed the complex bitmap images to be represented by vector based graphics and provided some advantages over the raster based graphics in applications, for example, where scalability is required. This paper presents an algorithmto convert bitmap images into SVG format. The algorithm is an integration of pixel-level triangulation, data dependent triangulation, a new image mesh simplification algorithm, and a polygonization process. Both triangulation techniques enable the image quality (especially the edge features) to be preserved well in the reconstructed image and the simplification and polygonization procedures reduce the size of the SVG file. Experiments confirm the effectiveness of the proposed algorithm.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2003년도 추계학술대회
• /
• pp.592-597
• /
• 2003

Thrust vector control using a coflow-counterflow concept is achieved by suction and blowing through a slot adjacent to a primary jet which is shrouded by a suction collar. In the present study, the flow characteristics of thrust vectoring is investigated using a numerical method. The nozzle has a design Mach number of 2.0, and the operation pressure ratio is varied to obtain various flow features of the nozzle flow. Test conditions are in the range of the nozzle pressure ratio from 6.0 to 10.0, and a suction pressure from 90kPa to 35kPa. Two-dimensional, compressible Navier-Stokes computations are conducted with RNG ${\kappa}-{\varepsilon}$ turbulence model. The computational results provide an understanding of the detailed physics of the thrust vectoring process. It is found that an increase in the nozzle pressure ratio leads to increased thrust efficiency but reduces the thrust vector angle.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2003년도 추계학술대회
• /
• pp.496-501
• /
• 2003

Computational study is performed to understand the fluidic thrust vectoring control of an axisymmetric nozzle, in which secondary gas injection is made in the divergent section of the nozzle. The nozzle has a design Mach number of 2.0, and the operation pressure ratio is varied to obtain the different flow features in the nozzle flow. The injection flow rate is varied by means of the injection port pressure. Test conditions are in the range of the nozzle pressure ratio from 3.0 to 8.26 and the injection pressure ratio from 0 to 1.0. The present computational results show that, for a given nozzle pressure ratio, an increase of the injection pressure ratio produces increased thrust vector angle, but decreases the thrust efficiency.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2002년도 전력전자학술대회 논문집
• /
• pp.293-296
• /
• 2002

This paper presents an implementation of Direct torque control(DTC) of Reluctance Synchronous Motor(RSM) drives for an industrial servo drive system with stator current space vector. The estimation of the stator flux magnitude are obtained by using the neural network from measuring the modulus and angle of the stator current space vector. The develolled digitally high-performance control system are shown a good response characteristic of control results and high performance features using 1.0kW RSM.

• 정보처리학회논문지B
• /
• 제10B권6호
• /
• pp.647-656
• /
• 2003

본 논문에서는 얼굴의 특징정보를 추출하여 사람의 얼굴과 유사한 캐리커처를 자동으로 생성해 주는 시스템을 제안한다. 본 논문에서 제안하는 시스템은 벡터를 기반으로 하기 때문에 사이즈에 제한 없이 자유로운 변형이 가능할 뿐만 아니라 2D 캐릭터에 자유로운 표정을 적용하는 데에도 쉽게 적용이 가능하다. 또, 벡터의 특징으로 인해 모바일 상에서도 적은 용량으로 이용가능하다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권11호
• /
• pp.4556-4572
• /
• 2015

Dynamic texture (DT) recognition is a challenging problem in numerous applications. In this study, we propose a new algorithm for DT recognition based on group sparsity structure in conjunction with chaotic feature vector. Bag-of-words model is used to represent each video as a histogram of the chaotic feature vector, which is proposed to capture self-similarity property of the pixel intensity series. The recognition problem is then cast to a group sparsity model, which can be efficiently optimized through alternating direction method of multiplier algorithm. Experimental results show that the proposed method exhibited the best performance among several well-known DT modeling techniques.