Earticle

This paper proposes a virtual driving system for enhancing efficient driving style. We compared fuel economies and levels of consumable wear on three driving simulation test (none, visual, visual+auditory). The test results show that higher fuel economy can be achieved and consumable wear be reduced when they received feedback from the proposed system compared to driving without eco-driving training.

Cloud computing provides a new way to build applications on on-demand infrastructures instead of building applications on fixed and rigid infrastructures. This paper discuss the methodology for cloud computing paradigm shift, that is, from the conventional computing to leveraging the features of cloud computing. This includes the way on how to redesign, build and deliver applications as well as the architectural challenges that must be taken for consideration when adopting and migrating into cloud computing.

In many real-world applications, data streams are usually collected in a decentralized manner such like sensor network, ubiquities sensor network, internet traffic analysis, and so on. In particularly, requirements for continuous, fast, high-volumes, adaptability, costly streaming data, an approximated analysis is needful for fast response to users on forward predicates. Distance estimate for both of “continuously” queries and streams is still a more challenge area because of a smaller or larger threshold selected is very easily to lead to a wrong result for continuously k-nearest neighbor queries. Therefore, we proposed a required filtering method to help to choose a well threshold of distance estimate in order to control error rates of approximated answers.

In this system we improved the performance of an automatic system for extracting leaf contour. The proposed leaf contour extraction method consists of three major procedures: the detection of four edge points, and contour tracing. Leaf detection includes two stages: feature extraction and matching. For the leaf contour extraction part, we present a new technique for automatically identifying the contour of a leaf in an image. For contour tracing, an Intelligent Scissor (IS) algorithm is applied. The color gradient magnitude and Canny edge detection are analyzed and included as the cost terms of the Intelligent Scissor algorithm. The color gradient magnitude cost term is implemented so that it can act directly on the three components of the color image. For the third procedure, we implement of the performance improvement. The main idea about proposed method is when the program detected the incorrect four edge points, we using the mouse and clicked the right position instead of the incorrect one.

The next era is foreseen to revolutionize the utility application with the technological advancements in IT and telecommunication. Intelligence will be seen in industrial machines equipped with embedded technologies enabling communication using low power wireless technology. Smart meter has recently gained lot of attention from the research community as this has raised unique challenges especially w.r.t security. Privacy preservation in such an environment is discussed to be critical as it may lead to devastating effects or even blackout which has been witnessed before. In this paper, we propose a privacy preserving algorithm which ensures resiliency against credentials theft with cryptographic operations where they are transported in a secure manner. We evaluated different encryption algorithms RC5, MD5, SHA1 in terms of computational and communication resources consumed by them in the proposed privacy preserving algorithm. In future, extensive evaluations and proofs of cryptographic algorithms will be included.

To optimize the search of the clonal selection algorithm for the optimal solution, the selection operator is improved and a novel clonal selection algorithm is designed. The selection operator is compared with such selection operators as Roulette Wheel Selection operator, Tournament Selection operator, and Sampling Selection operator. The selection operator is improved to increase the accuracy and efficiency for finding the optimal solutions. The experimental results show that the novel clonal selection algorithm performs better than the traditional clonal selection algorithm due to the adjustment of the selection operator.

SSDs use multiple NAND flash memory chips as storage media and deploy large sized RAM inside it in order to maintain the FTL mapping table. The rest portion of the inner RAM can be used as buffer. The buffer absorbs the read/write requests by file systems and thus the resulting write requests to NAND flash memory is determined by the buffer replacement scheme. The block-level LRU replacement schemes, which manages the buffer in NAND block unit, generates a large sized write pattern that is NAND-friendly. However, the existing schemes do not consider the re-write probability of each page. This paper presents a new block-level replacement policy for SSDs. The presented scheme evicts only cold pages that its re-write probability is low considering the re-write probability, which can contribute to improve the buffer hit ratio.

In this paper, we explain a study on natural user interface (NUI) in human gesture recognition using RGB color information and depth information by Kinect camera from Microsoft Corporation. To achieve the goal, hand tracking and gesture recognition have no major dependencies of the work environment, lighting or users’ skin color, libraries of particular use for natural interaction and Kinect device, which serves to provide RGB images of the environment and the depth map of the scene were used. An improved Camshift tracking algorithm combined with depth information is used to tracking hand motion, and then an associative method of HMM and FNN is propose for gesture recognition step. The experimental results show out its good performance and it has higher stability and accuracy as well.

We propose a watermarking scheme for the authentication of binary images. We transform the image using the binary wavelet transform and partition the resulting coefficients into tree-structured blocks. The size of each block is decided by the number of flippable coefficients in the block. The flippable coefficients are found in the frequency domain by investigating high frequency coefficients. A random bit sequence generated by the block identifying key is embedded in the flippable coefficients in each block. Block-wise localization of tampered regions is accomplished with imperceptible visual degradations. Counter-measures against the vector quantization attack to which most existing watermarking schemes are vulnerable may also be implemented in our scheme

We proposed a method complementing failure of combining DNA fragments, defect of conventional contig assembly programs. In the proposed method, very long DNA sequence data are made into a prototype of fragment of about 700 bases that can be analyzed by automatic sequence analyzer at one time, and then matching ratio is calculated by comparing a standard prototype with 3 fragmented clones of about 700 bases generated by the PCR method. In this process, the time for calculation of matching ratio is reduced by Compute Agreement algorithm. Two candidates of combined fragments of every prototype are extracted by the degree of overlapping of calculated fragment pairs, and then degree of combination is decided using a fuzzy inference method that utilizes the matching ratios of each extracted fragment, and A, C, G, T membership degrees of each DNA sequence, and previous frequencies of each A, C, G, T. In this paper, DNA sequence combination is completed by the iteration of the process to combine decided optimal test fragments until no fragment remains. For the experiments, fragments of about 700 bases were generated from each sequence of 10,000 bases and 100,000 bases extracted from ‘PCC6803’, complete protein genome. From the experiments by applying random mutations on these fragments, we could see that the proposed method was faster than FAP program, and combination failure, defect of conventional contig assembly programs, did not occur.

In this paper, a novel image zooming scheme for indexed color images is proposed. In the proposed scheme, the bilinear interpolation technique is employed to enlarge the original image. Then, the pixel grouping process is executed to classify the pixels in the enlarged image. From the experimental results, it is shown that the proposed scheme provides better image quality than the pixel copy technique for the zooming of the indexed color images.

In this paper, we have proposed a method that consists of combination of different methods. First we have performed enhancement on breast mammogram to enhance the image quality. After that discrete cosine transform has been applied for features extraction. Bayesian Classifier has been used for classification into benign and malignant. It has been noted that results are very much satisfactory. We have used MIAS data set for experimentation purpose. Proposed method performs good when we have tested on different images.

As the significant role of the software platform in the smart phone is now propagating beyond the mobile industry, the software platform is evolving to provide unified multi-screen service facilities over different types of devices such as digital TV and Set-Top Box (STB). However, the application-specific hardware architectures impose difficulties for providing unified software platform for heterogeneous devices. This paper analyzes the differences between mobile and TV/STB System-On-Chip (SoC) architecture and proposes a unified hardware abstraction layer (HAL) to support integration of the software platform on different devices. For the evaluation and demonstration of the proposed design, the unified HAL is applied to the mobile Android platform to implement multi-screen services.

In on-cell type touch panel structure, screen printing technology which uses Ag paste on ITO film is one of the general processes for Bus bar formation. Considering the growing of capacitance touch panel industry and slim design trend, decreasing the Bezel width is necessary. The high technology should be developed for Bus bar printing. Thus, Bus bar formation process was designed by metal deposition and photo lithography in this theory. By applying multiple layers structure process, the comparisons of adhesion strength between ITO films, resistances and pitch width were proceeded. Based on the comparisons, Mo/Al/Mo structure was known as the best for fine pitch bus bar formation process.

Recently, in-network aggregate query processing techniques have problems such as high energy consumption in sensor nodes, low accuracy of query processing results, and long query processing time. In order to solve these problems and to enhance the efficiency of aggregate query processing in wireless sensor networks, this paper proposes In-network Query Method(IQM). IQM divides a query region into several cells according to the distribution of sensor nodes and builds a Quad-tree, and then processes aggregate queries in parallel for each cell region according to routing.

Various types of data structures and algorithms have been proposed to extract frequently occurring patterns from a given data set. In particular, several tree structures have been devised to represent the input data set for efficient pattern discovery. One of the fastest frequent pattern mining algorithms known to date is the CATS algorithm, which can efficiently represent the whole data set and allow mining with a single scan over the database. In this paper, we propose an efficient tree structure and its associated algorithm that provides a considerable performance improvement over CATS in terms of memory usage and processing time. We have demonstrated the effectiveness of our algorithm and performance improvement over the existing approach by a series of experiments.

This paper presents a PC-based thermal cycler with a local-host architecture, where a PC is employed as the host. The local system only executes the thermal cycling process and the PC is responsible for the management of PCR protocols with a convenient GUI. The presented local-host system greatly reduces the development time and the maintenance cost, providing an economical PCR thermal cycler.

The polymerase chain reaction (PCR) is a method utilized in most of the experiments handling genetic materials. PCR amplifies the target genetic material that is to be analyzed. This paper demonstrates the accuracy of the temperature measured by the thermistor. Since the thermistor is attached to the bottom of the micro PCR chip, there becomes a gap between the chamber and the thermistor. This experiment measures the temperature difference due to the gap. Micro PCR chip has a chamber made of double-sided tape and OHP film integrated upon a PCB substrate printed with a heater pattern. The thermistor to measure the chamber temperature is attached to the bottom of the substrate. In the experiment, a thermo-couple was inserted into the chamber directly to measure the actual temperature. The results showed no significant difference between the actual chamber temperature measured by the thermo-couple and the temperature measured by the thermistor attached at the bottom of the substrate.

3D image technology advancement served as the turning point that shows the potential of the 3D technology to be used for next generation broadcasting technology. Due to this reason, diverse image contents were produced using 3D technology, but side-effects began to surface while watching 3D image contents. Some users complained of fatigue, dizziness, and nausea while viewing the contents produced in 3D. As such, researches are underway on the 3D syndrome symptoms detection. However, the reality is that there are not many researches that address basic measures and the technologies themselves. Accordingly, this research paper analyzed the correlation of the brain wave signals at the frontal lobe/occipital lobe parts to analyze the change in the brain wave signal during 3D image viewing when 3D syndrome occurs and to realize the automatic 3D syndrome detection system with the key findings.

This thesis investigates how KoShik, otherwise known as the talking elephant, can have a human-like voice articulation. The mechanism of articulating a human sound by an elephant can be described as a unique excitation process in which the elephant first places its nose in its mouth, blows air into the mouth, produces the basic frequency by pressing the nose with its mouth and teeth, and then articulates the sounds through the narrow airway between the inner skin and soft palate of its mouth. This vocalization process is similar to that of its breeder with respect to the basic pitch and 1-2 formant frequency. The talking elephant KoShik is different from other so-called talking animals in its uniqueness of making a sound by utilizing the nose, more like playing a musical instrument than just mimicking human sounds, as well as of its activeness to talk.

In H.264 video compression standard, B frame bi-directional predictions and its DIRECT mode coding are included to enhance video compression efficiency. However, such tools require a complex mode decision process and result in a long computation time. Therefore, fast algorithms are crucial in implementing H.264. In this paper, to save B frame encoding time, a fast DIRECT mode prediction algorithm is proposed, which determines DIRECT mode at an early stage by optimal mode pattern matching. Simulations show that the proposed algorithm can determine DIRECT mode without a complex mode decision process for 42% more macroblocks and, on mis-determination for DIRECT mode, cut down its rate-distortion (RD) cost increment by half, compared with Jin’s algorithm. This enables a fast B frame encoding with less quality degradation.

As audio players become smaller in size and more portable, users can play and listen to music anywhere. More people are using headphones outdoors, and this leads to a higher frequency of traffic accidents. In this paper, the study proposes an architecture of automatic warning system on urgent traffic situation for headphone users. This study proposes a method of warning pedestrians wearing headphones so that they can quickly respond to external sounds and detect an approaching accident. The system was evaluated to be effective in processing living noises and car horn sounds through the simulation of an urgent traffic situation.

In this article, we propose an utterance verification technique for keyword spotting. The keyword spotting system analyzes a given spoken content and searches every speech segment in which one of pre-defined keywords is uttered. To maintain a stable recognition performance in the system, we propose an utterance verification technique that verifies whether a found utterance, or a candidate keyword segment, can be categorized as a keyword. The proposed approach employs a confidence measure based on the recognition results (N-best log-likelihood). In keyword spotting experiments using spoken broadcast news, our approach achieved superior performance compared to the conventional approach.

To break through the energy-latency tradeoff caused by duty cycling in the existing wireless sensor network (WSN), in this paper we propose a new RF wakeup sensor, which is a dedicated small RF module to check potential communications by sensing the presence of a RF signal. With RF wakeup sensor each node no longer requires duty cycling, eliminating both sleep delay and idle listening. The distinctive feature of our RF design from the existing radio sensor studies is that the RF wakeup sensor can provide the same sensitivity as the underlying RF communication module with two orders of magnitude less energy.

In this paper, we propose a 3D coordinate estimation for 3D space using a mono camera in mobile devices. Generally, a marker is used for the detection of 3D movement in 2D images. However, this marker-based system has obvious defects since markers always need to be included in the image. To overcome this problem, we replaced the markers with a planar face by estimating the face detection. The experimental result indicates the feasibility of the proposed algorithm for face movement-based mobile user interfaces.

The bandwidth adaptation is the technique that allows the flexibility in bandwidth allocation for a call. Using the bandwidth adaptation technique, the number of call admission in the system can be increased significantly. In this paper we propose a priority based bandwidth adaptation scheme that can release multi-level of bandwidth from the existing calls to accept call requests. The amount of released bandwidth is based on the number of existing bandwidth adaptive calls and the priority of requesting traffic call. This priority scheme does not reduce the bandwidth utilization. Moreover, the proposed bandwidth adaptation strategy provides significantly reduced call blocking probability for the higher priority traffic calls. The performance analyses show the improvement of the proposed scheme.

In this paper we use Ubiquitous Sensor Network(USN) and image information to develop remote navigation system for mobile robot. USN with ultrasonic sensors is used to estimate local position of mobile robot. Using this information of position, mobile robot can navigate in certain area autonomously. And USB web-camera is used to obtain real-time images of remote places and monitor it, when mobile robot navigates remotely.

Separating technique for singing voice from music accompaniment is very useful in original sound type Karaoke instrument. We propose a real-time system to separate singing voice from music accompaniment for stereo recordings. Proposed algorithm consists of two stages. The first stage is a spectral change detector. The last stage is a selective vocal separation in frequency bins. Our system consists of a DSP board and a multichannel audio interface board. The DSP board is a TMS320C6713 DSK. And multichannel audio interface has six (three stereos) channels. The proposed vocal removal algorithms are embedded in DSP. Listening tests with extracted MR from proposed system show vocal separating and removal tasks successfully.

In this paper, we proposed Cost Effective Acknowledgement (CEA) method which was based on Multiple Acknowledgement (MA) [1] method and proposed by Smart Block Medium Access Control (SBMAC) [2]. The proposed method can reply the information of transmission error states of all Senders within smaller storage space. Especially, the number of transmission and the frame length was minimized to reduce transmission error. The performance of the proposed method and conventional methods such as Normal Multiple Acknowledgement (NMA) and Selective Multiple Acknowledgement (SMA) method [3] was calculated and compared with each other. The calculation results showed the best performance in case of the proposed method.

Datamining is interested and applied in several areas around us. In other words, it is to discover hidden useful correlations and to predict future by extracting the process to make decisions in actionable information. A variety of data can be converted to interpret in other format. Even a simple conversion can find a big difference in analyzing the results. In this paper, the three principal components for analysis and changes were used Random projections and data conversion techniques. The correct classification is compared by performance of two methods for viewing the sample data set using the Bayesian algorithm.