• JSTS:Journal of Semiconductor Technology and Science
• /
• v.13 no.2
• /
• pp.157-169
• /
• 2013

Scale Invariant Feature Transform (SIFT) generates image features widely used to match objects in different images. Previous work on hardware-based SIFT implementation requires excessive internal memory and hardware logic [1]. In this paper, a new hardware organization is proposed to implement SIFT with less memory and hardware cost than the previous work. To this end, a parallel Gaussian filter bank is adopted to eliminate the buffers that store intermediate results because parallel operations allow all intermediate results available at the same time. Furthermore, the processing order is changed from the raster-scan order to the block-by-block order so that the line buffer size storing the source image is also reduced. These techniques trade the reduction of memory size with a slight increase of the execution time and external memory bandwidth. As a result, the memory size is reduced by 94.4%. The proposed hardware for SIFT implementation includes the Descriptor generation block, which is omitted in the previous work [1]. The addition of the hardwired descriptor generation improves the computation speed by about 30 times when compared with the previous work.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.10a
• /
• pp.252-255
• /
• 2010

In this thesis, it suggests large scale interactive display system which is able to various touch interaction and bases on infrared LED BAR and using 3D. Interaction layer formed on space from screen which is able to feel 3D using suggested IR LED BAR. It gets the image in real time what is composed in interaction section using infrared camera with band pass filter. The image finds touch interaction coordinate through image processing module and saves as packet. It send packet to server through network data communication. It analyze packet by metaphor analysis module and save as metaphor event and send it to contents. On contents, it practices to metaphor event result in real time so it makes use touch interaction in stereopsis. According to this process, it does not need touch the screen at firsthand but it is possible system and touch interaction so touch interaction is possible while use 3D.

• Journal of the Institute of Convergence Signal Processing
• /
• v.9 no.2
• /
• pp.121-128
• /
• 2008

Accurate voice activity detection have a great impact on performance of speech applications including speech recognition, speech coding, and speech communication. In this paper, we propose methods for voice activity detection that can adapt to various car noise situations during driving. Existing voice activity detection used various method such as time energy, frequency energy, zero crossing rate, and spectral entropy that have a weak point of rapid. decline performance in noisy environments. In this paper, the approach is based on existing spectral entropy for VAD that we propose voice activity detection method using MFB(Met-frequency filter banks) spectral entropy, gradient FFT(Fast Fourier Transform) spectral entropy. and gradient MFB spectral entropy. FFT multiplied by Mel-scale is MFB and Mel-scale is non linear scale when human sound perception reflects characteristic of speech. Proposed MFB spectral entropy method clearly improve the ability to discriminate between speech and non-speech for various in noisy car environments that achieves 93.21% accuracy as a result of experiments. Compared to the spectral entropy method, the proposed voice activity detection gives an average improvement in the correct detection rate of more than 3.2%.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2006.11a
• /
• pp.89-92
• /
• 2006

본 논문은 SIFT(Scale Invariant Feature Transform)알고리즘의 실시간처리 응용프로그램 작성기법을 기술하고 있는데, 단일 프로세서에서 병렬처리 기능을 지원하도록 설계된 SSE 명령어 집합을 사용하여 가우시안 convolution을 구현하고 있다. SIFT알고리즘의 Scale-space를 생성하는 과정에 수행되는 가우시안 Convolution은 연산시간이 과도하게 요구된다.[1] 2D의 가우시안 필터가 영상을 구성하는 모든 셀과 1:1로 연산을 수행하므로 이 연산의 소요시간은 영상의 가로, 세로 길이 그리고 필터의 크기에 비례하여 결정된다. 이 논문에서 제안하는 방법은 연산을 위해 CPU 내부로 한번 읽어 들인 픽셀자료에 대해 가능한 모든 연산을 SSE 명령어 집합을 사용하여 수행함으로써 병렬 연산에 의한 연산시간 절감과 메모리 접근 최소화를 통한 입출력시간 절감을 통해 전체 연산시간을 단축 하였다.

• Journal of Korea Multimedia Society
• /
• v.15 no.3
• /
• pp.336-344
• /
• 2012

As an important part of the Computer Vision, the object recognition and tracking function has infinite possibilities range from motion recognition to aerospace applications. One of methods to improve accuracy of the object recognition, are uses colors which have robustness of orientation, scale and occlusion. Computational cost for extracting features can be reduced by using color. Also, for fast object recognition, predicting the location of the object recognition in a smaller area is more effective than lowering accuracy of the algorithm. In this paper, we propose a method that uses SURF descriptors which applied with color model for improving recognition accuracy and combines with Kalman filter which is Motion estimation algorithm for fast object tracking. As a result, the proposed method classified objects which have same patterns with different colors and showed fast tracking results by performing recognition in ROI which estimates future motion of an object.

• Journal of Korean Society on Water Environment
• /
• v.23 no.5
• /
• pp.642-649
• /
• 2007

The effectiveness of add-on tertiary treatment processes for the polishing of the effluent of a biological nutrient removal (BNR) system from a modified $A^2/O$ process has been examined under the field condition with pilot-scale plants. The add-on treatment processes of 1) combined biofilm anoxic reactor and sand filtration, and 2) two-stage denitrification filter had been operated with various operating conditions. The experimental results indicated that two-stage denitrification filter could produced a better polished tertiary effluent. Filtration rate of $150m^3/m^2{\cdot}d$ for the 2-stage denitrifying filter could decrease the nitrate removal probably due to shorter detention time that caused insufficient reaction for denitrification. Two stage denitrification filter operated with M/N ratio of 3.0 and filtration rate of $100m^3/m^2{\cdot}d$ produced the tertiary effluent with nitrate and SS concentraitons of 2.8 mg/L and 2.3 mg/L, respectively. When the operating temperature reduced $30^{\circ}C$ to $18^{\circ}C$, $NO_3{^-}-N$ removal efficiency decreased from 73% to 68%.

• Bulletin of the Korean Chemical Society
• /
• v.25 no.12
• /
• pp.1829-1832
• /
• 2004

We have developed a new femtosecond probe technique by using stimulated Raman spectroscopy. The cross phase modulation in femtosecond time scale associated with off-resonant interaction was shown to be eliminated by integrating the transient gain/loss signal over the time delay between the Raman pump pulse and the continuum pulse. The stimulated Raman gain of neat cyclohexane was obtained to demonstrate the feasibility of the technique. Spectral and temporal widths of stimulated Raman spectra were controlled by using a narrow band pass filter. Femtosecond stimulated Raman spectroscopy was proposed as a highly useful probe in time-resolved vibrational spectroscopy.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.2 no.3
• /
• pp.39-52
• /
• 2009

Scale Invariant Feature Transform (SIFT) is an effective algorithm in object recognition, panorama stitching, and image matching, however, due to its complexity, real time processing is difficult to achieve with software approaches. This paper proposes using a reconfigurable hardware processor with integer half kernel. The integer half kernel Gaussian reduces the Gaussian pyramid complexity in about half [] and the reconfigurable processor carries out a parallel implementation of a full search Fast SIFT algorithm. We use a low memory, fine grain single instruction stream multiple data stream (SIMD) pixel processor that is currently being developed. This implementation fully exposes the available parallelism of the SIFT algorithm process and exploits the processing and I/O capabilities of the processor which results in a system that can perform real time image and video compression. We apply this novel implementation to images and measure the effectiveness. Experimental simulation results indicate that the proposed implementation is capable of real time applications.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.615-619
• /
• 2009

A real-time, continuous-scale image interpolation method is proposed based on bi-linear interpolation with directionally adaptive low-pass filtering. The proposed algorithm has been optimized for hardware implementation. The original bi-linear interpolation method has blocking artifact. The proposed algorithm solves this problem using directionally adaptive low-pass filtering. It can also solve the severely problem by selection choosing low-pass filter coefficients. Therefore the proposed interpolation algorithm can realize a high-quality image scaler for various imaging systems, such as digital camera, CCTV and digital flat panel display, to name a few.

• Journal of Environmental Science International
• /
• v.11 no.8
• /
• pp.819-827
• /
• 2002

Laboratory scale experiments were conducted to evaluate the performance of a biofilter for eliminating dimethyl sulfide(DMS). A commercial compost/pine bark nugget mixture served as the biofilter material for the experiments. The gas flow rate and DMS concentration entering the filter were varied to study their effect on the biofilter efficiency. The operating parameters, such as the residence time, inlet concentration, pH, water content, and temperature, were all monitored throughout the filter operation. The kinetic dependence of the DMS removal along the column length was also studied to obtain a quantitative description of the DMS elimination. High DMS removal efficiencies(>95％) were obtained using the compost filter material seeded with activated sludge. DMS pollutant loading rates of up to 5.2 and 5.5 g-DMS/m$^3$/hr were effectively handled by the upflow and downflow biofilter columns, respectively. The macrokinetics of the DMS removal were found to be fractional-order diffusion-limited over the 9 to 25 ppm range of inlet concentrations tested. The upflow column had an average macrokinetic coefficient(K$\_$f/) of 0.0789 $\pm$ 0.0178 ppm$\^$$\sfrac{1}{2}$//sec, while the downflow column had an average coefficient of 0.0935 $\pm$ 0.0200 ppm$\^$$\sfrac{1}{2}$//sec. Shorter residence times resulted in a lower mass transfer of the pollutant from the gas phase to the aqueous liquid phase, thereby decreasing the efficiency.