• International journal of advanced smart convergence
• /
• 제12권4호
• /
• pp.171-176
• /
• 2023

This study is about a speech recognition error correction system designed to detect and correct speech recognition errors before natural language processing to increase the success rate of intent analysis in natural language processing with optimal efficiency in various service domains. An encoder is constructed to embedded the correct speech token and one or more error speech tokens corresponding to the correct speech token so that they are all located in a dense vector space for each correct token with similar vector values. One or more utterance tokens within a preset Manhattan distance based on the correct utterance token in the dense vector space for each embedded correct utterance token are detected through an error detector, and the correct answer closest to the detected error utterance token is based on the Manhattan distance. Errors are corrected by extracting the utterance token as the correct answer.

• 대한임베디드공학회논문지
• /
• 제3권2호
• /
• pp.102-108
• /
• 2008

This paper describes a mapping and localization based on wave algorithm[11] and Kalman filter for effective SLAM. Each robot in a multi robot system has its own task such as building a map for its local position. By combining their data into a shared map, the robot scans actively seek to verify their relative locations. For simultaneous localization the algorithm which is well known as Kalman Filter (KF) is used. For modelling the robot position we wish to know three parameters (x, y coordinates and its orientation) which can be combined into a vector called a state variable vector. The Kalman Filter is a smart way to integrate measurement data into an estimate by recognizing that measurements are noisy and that sometimes they should ignored or have only a small effect on the state estimate. In addition to an estimate of the state variable vector, the algorithm provides an estimate of the state variable vector uncertainty i.e. how confident the estimate is, given the value for the amount of error in it.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅳ
• /
• pp.2212-2215
• /
• 2003

In this paper, we proposed speech feature vector extractor for embedded system using TMS 320C31 DSP chip. For this extractor, we used algorithm using cepstrum coefficient based on LPC(Linear Predictive Coding) that is reliable algorithm to be is widely used for speech recognition. This system extract the speech feature vector in real time, so is used the mobile system, such as cellular phones, PDA, electronic note, and so on, implemented speech recognition.

• 대한임베디드공학회논문지
• /
• 제9권3호
• /
• pp.173-182
• /
• 2014

The CRT monitor display system for SONAR signal that are commonly used in ships or naval vessels uses vector scanning method. Therefore the processing circuits of the system are complex. Also the purchase of parts is difficult as well as high-cost because the production had been shut down. FPGA-based embedded system is flexible to various digital applications because it can be able to simplify processing circuits and to make a easy customized design for end user, and it provides low-cost high-speed performance. In this paper, we describe an implementation of FPGA embedded system for real-time SONAR signal display using the triple buffering method to overcome some weakness of existing CRT system. Our system provides real-time acquisition and display capability of SONAR signal, and removes afterimage effect that is a critical problem of the system proposed in the preceding study.

• 한국정보통신학회논문지
• /
• 제10권7호
• /
• pp.1222-1228
• /
• 2006

본 논문에서는 적응적 임계값과 반화소 움직임 벡터를 이용한 비디오 워터마킹 방법을 제안한다. 기존의 방법은 모든 매크로블록에 대해 워터마크삽입 조건을 검색하고, 임의의 움직임 벡터에 워터마크가 삽입되므로 움직임 벡터 변경이 많고, 비가시성이 떨어진다. 제안한 방법은 PMV를 이용하여 임계값을 생성하고, 임계값보다 큰 움직임 벡터에 워터마크를삽입하므로 움직임 벡터 변경 횟수가 적고, 비가시성이 향상된다. 실험결과 제안한 방법은 바디오 비트스트림에 따라 평균 5.4 dB ~ 9.3 dB 의 비가시성 특성이 향상됨을 알 수 있었다.

• 대한임베디드공학회논문지
• /
• 제9권1호
• /
• pp.17-24
• /
• 2014

In this paper, we implement and evaluate the performance of a vector-based rasterization algorithm for 3D graphics by using a SIMD (single instruction multiple data) many-core processor architecture. In addition, we evaluate the impact of a data-per-processing elements (DPE) ratio that is defined as the amount of data directly mapped to each processing element (PE) within many-core in terms of performance, energy efficiency, and area efficiency. For the experiment, we utilize seven different PE configurations by varying the DPE ratio (or the number PEs), which are implemented in the same 130 nm CMOS technology with a 500 MHz clock frequency. Experimental results indicate that the optimal PE configuration is achieved as the DPE ratio is in the range from 16,384 to 256 (or the number of PEs is in the range from 16 and 1,024), which meets the requirements of mobile devices in terms of the optimal performance and efficiency.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권8호
• /
• pp.578-584
• /
• 2004

The CSVM(Current Support Vector Machine) that is a digital architecture performing all phases of recognition process including kernel computing, learning, and recall of SVM(Support Vector Machine) on a chip is proposed. Concurrent operation by parallel architecture of elements generates high speed and throughput. The classification problems of bio data having high dimension are solved fast and easily using the CSVM. Quadratic programming in original SVM learning algorithm is not suitable for hardware implementation, due to its complexity and large memory consumption. Hardware-friendly SVM learning algorithms, kernel adatron and kernel perceptron, are embedded on a chip. Experiments on fixed-point algorithm having quantization error are performed and their results are compared with floating-point algorithm. CSVM implemented on FPGA chip generates fast and accurate results on high dimensional cancer data.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.385-387
• /
• 2004

In this paper, we have designed a fingerprint cognition system based on the embedded Linux. The proposed algorithm in this paper use the wavelet transform to derive the special feature vector from the captured fingerprint and a probabilistic neural network is used to compare the feature vectors for the fingerprints. The system consists of server PC based on the Linux and the client based on the embedded Linux. The client is a Tynux box-x board using the PXA-255 CPU. For the acquisition of the fingerprint image, we use d the AS-S2 semiconductor sensor. The system is likely to be used to develop a police inspection system.

• 한국음향학회지
• /
• 제30권8호
• /
• pp.461-467
• /
• 2011

제한된 대역폭을 효율적으로 사용하기 위해서 도입된 가변 전송률은 먼저 신호의 정확한 분류를 필요로 한다. 특히 멀티미디어 서비스가 보편화 되면서 음성/음악 신호 분류의 중요성도 높아지게 되었다. 음성/음악 분류기 중, 서포트벡터머신 (SVM)을 이용한 분류기는 높은 분류 정확도로 주목받고 있다. 그러나 SVM는 많은 계산량과 저장 공간을 요구하므로 효율적인 구현이 요구되며, 특히 임베디드 시스템과 같이 자원이 제한 적인 경우에는 더욱 그러하다. 본 논문에서는 먼저 SVM을 이용한 음성/음악 분류기의 임베디드 시스템으로의 구현을 실행시간과 에너지소비의 관점에서 분석하고, 효율적인 구현을 위한 두가지 방법들을 제안한다. 서포트벡터의 판별결과에의 기여도를 바탕으로 기여도가 낮은 벡터들을 제외하는 방법과, 음성/음악 신호에 기본적으로 존재하는 각 프레임간의 상관관계를 이용하여 입력신호의 일부를 건너뛰는 방법이다. 이 기법들은 SVM의 학습 시 사용되는 다른 최적화 기법에 관계없이 적용이 가능하며, 실험을 통해 분류의 정확도, 실행시간, 그리고 에너지소비의 관점에서 그 성능을 증명하였다.

• 한국정보통신학회논문지
• /
• 제24권7호
• /
• pp.956-962
• /
• 2020

최근 임베디드 시스템에서 사용되는 데이터의 크기가 증가함에 따라, 대용량의 데이터를 안전하게 수신하기 위한 ECC (Error Correction Code) 복호화 연산의 필요성이 강조되고 있다. 본 논문에서는 GPU가 내장된 임베디드 시스템에서 해밍 코드를 사용하여 ECC 복호화를 할 때, 신드롬 벡터를 계산하는 연산의 수행을 가속할 방법을 제안한다. 제안하는 가속화 방법은, 복호화 연산의 행렬-벡터 곱셈이 희소 행렬을 나타내는 자료 구조 중 하나인 CSR (Compressed Sparse Row) 형식을 사용하고, GPU의 CUDA 커널에서 병렬적으로 수행되도록 한다. 본 논문에서는 GPU가 내장된 실제 임베디드 보드를 사용하여 제안하는 방법을 검증하였고, 결과는 GPU 기반으로 가속된 ECC 복호화 연산이 CPU만을 사용한 경우에 비하여 수행 시간이 감소하는 것을 보여준다.