We describe a technique for the registration of three dimensional (3D) knee femur surface points from MR image data sets; it is a technique that can track local cartilage thickness changes over time. In the first coarse registration step, we use the direction vectors of the volume given by the cloud of points of the MR image to correct for different knee joint positions and orientations in the MR scanner. In the second fine registration step, we propose a global search algorithm that simultaneously determines the optimal transformation parameters and point correspondences through searching a six dimensional space of Euclidean motion vectors (translation and rotation). The present algorithm is grounded on a mathematical theory - Lipschitz optimization. Compared with the other three registration approaches (ICP, EM-ICP, and genetic algorithms), the proposed method achieved the highest registration accuracy on both animal and clinical data.

We propose a blind dereverberation method based on spectral subtraction using a multi-channel least mean squares (MCLMS) algorithm for distant-talking speech recognition. In a distant-talking environment, the channel impulse response is longer than the short-term spectral analysis window. By treating the late reverberation as additive noise, a noise reduction technique based on spectral subtraction was proposed to estimate the power spectrum of the clean speech using power spectra of the distorted speech and the unknown impulse responses. To estimate the power spectra of the impulse responses, a variable step-size unconstrained MCLMS (VSS-UMCLMS) algorithm for identifying the impulse responses in a time domain is extended to a frequency domain. To reduce the effect of the estimation error of the channel impulse response, we normalize the early reverberation by cepstral mean normalization (CMN) instead of spectral subtraction using the estimated impulse response. Furthermore, our proposed method is combined with conventional delay-and-sum beamforming. We conducted recognition experiments on a distorted speech signal simulated by convolving multi-channel impulse responses with clean speech. The proposed method achieved a relative error reduction rate of 22.4% in relation to conventional CMN. By combining the proposed method with beamforming, a relative error reduction rate of 24.5% in relation to the conventional CMN with beamforming was achieved using only an isolated word (with duration of about 0.6 s) to estimate the spectrum of the impulse response.

The increasing demand of low power Direct Digital Frequency Synthesizer (DDFS) leads to the requirement of efficient compression methods to reduce ROM size for storing sine function values. This paper presents a technique to achieve very high compression ratio by using the optimized four-segment linear difference method. The proposed technique results in the ROM compression ratio of about 117.3:1 and the word size reduction of 6 bits for the design of a DDFS with 11-bit sine amplitude output. This high compression ratio result is very promising to meet the requirement of low power consumption and low hardware complexity in digital VLSI technology.

In the text summarization field, a table-of-contents is a type of indicative summary that is especially suited for locating information in a long document, or a set of documents. It is also a useful summary for a reader to quickly get an overview of the entire contents. The current models for generating a table-of-contents produced relatively low quality output with many meaningless titles, or titles that have no overlapping meaning with the corresponding contents. This problem may be due to the lack of semantic information and topic information in those models. In this research, we propose to integrate supportive knowledge into the learning models to improve the quality of titles in a generated table-of-contents. The supportive knowledge is derived from a hierarchical clustering of words, which is built from a large collection of raw text, and a topic model, which is directly estimated from the training data. The relatively good results of the experiments showed that the semantic and topic information supplied by supportive knowledge have good effects on title generation, and therefore, they help to improve the quality of the generated table-of-contents.

XQuery has become the standard for querying XML. Just like SQL, XQuery allows nested expressions. To optimize XQuery processing, a lot of research has been done on normalization, i.e., transforming nested expressions to equivalent unnested ones. Previous normalization rules are classified into two categories—source-level/ and algebra-level/—depending on whether a construct is specified in the XQuery syntax or as equivalent algebraic expressions. From an implementation point of view, the former is preferable to the latter since it can be implemented in a variety of XQuery engines with different algebras. However, existing source-level rules have several problems: They do not handle quantified expressions, incur duplicated query results, and use many temporary files. In this paper, we propose new source-level normalization rules that solve these problems. Through analysis and experiments, we show that our normalization rules can reduce query execution time from hours to a few seconds and can be adapted to a variety of XQuery engines.

With the emergence of Web 2.0, social tagging systems become highly popular in recent years and thus form the so-called folksonomies. Personalized tag recommendation in social tagging systems is to provide a user with a ranked list of tags for a specific resource that best serves the user's needs. Many existing tag recommendation approaches assume that users are independent and identically distributed. This assumption ignores the social relations between users, which are increasingly popular nowadays. In this paper, we investigate the role of social relations in the task of tag recommendation and propose a personalized collaborative filtering algorithm. In addition to the social annotations made by collaborative users, we inject the social relations between users and the content similarities between resources into a graph representation of folksonomies. To fully explore the structure of this graph, instead of computing similarities between objects using feature vectors, we exploit the method of random-walk computation of similarities, which furthermore enable us to model a user's tag preferences with the similarities between the user and all the tags. We combine both the collaborative information and the tag preferences to recommend personalized tags to users. We conduct experiments on a dataset collected from a real-world system. The results of comparative experiments show that the proposed algorithm outperforms state-of-the-art tag recommendation algorithms in terms of prediction quality measured by precision, recall and NDCG.

Due to the increased preservation efforts, large amounts of past Web data have been stored in Web archives and other archival repositories. Utilizing this data can offer certain benefits to users, for example, it can facilitate page understanding. In this paper, we propose a system for interactive exploration of page histories. We demonstrate an application called Page History Explorer (PHE) for summarizing and visualizing histories of Web pages. PHE portrays the overview of page evolution, characterizes its typical content over time and lets users observe page histories from different viewpoints. In addition, it enables flexible comparison of histories of different pages.

AC-waveform synthesis with quantum-mechanical accuracy has been attracting many researchers, especially metrologists in national metrology institutes, not only for its scientific interest but its potential benefit to industries. We describe the current status at National Metrology Institute of Japan of development of a Josephson arbitrary waveform synthesizer based on programmable and pulse-driven Josephson junction arrays.

Recently, overlap frequency-domain equalization (FDE) based on minimum mean square error (MMSE) criterion which requires no guard interval (GI) insertion was proposed for signal transmission using multi-code direct sequence code division multiple access (DS-CDMA) in a frequency-selective fading channel. Another promising equalization technique is time-domain sliding-window chip equalization (SWCE). In this paper, the bit error rate (BER) performances achievable with overlap FDE and SWCE are compared. It is shown that, by extending the fast Fourier transform (FFT) block size, overlap FDE can achieve almost the same BER performance as SWCE with much less computational complexity than SWCE.

One typical phonotactic system for language recognition is parallel phone recognition followed by vector space modeling (PPRVSM). In this system, various phone recognizers are applied in parallel and fused at the score level. Each phone recognizer is trained for a known language, which is assumed to extract complementary information for effective fusion. But this method is limited by the large amount of training samples for which word or phone level transcription is required. Also, score fusion is not the optimal method as fusion at the feature or model level will retain more information than at the score level. This paper presents a new strategy to build and fuse parallel phone recognizers (PPR). This is achieved by training multiple acoustic diversified phone recognizers and fusing at the feature level. The phone recognizers are trained on the same speech data but using different acoustic features and model training techniques. For the acoustic features, Mel-frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) are both employed. In addition, a new time-frequency cepstrum (TFC) feature is proposed to extract complementary acoustic information. For the model training, we examine the use of the maximum likelihood and feature minimum phone error methods to train complementary acoustic models. In this study, we fuse phonotactic features of the acoustic diversified phone recognizers using a simple linear fusion method to build the PPRVSM system. A novel logistic regression optimized weighting (LROW) approach is introduced for fusion factor optimization. The experimental results show that fusion at the feature level is more effective than at the score level. And the proposed system is competitive with the traditional PPRVSM. Finally, the two systems are combined for further improvement. The best performing system reported in this paper achieves an equal error rate (EER) of 1.24%, 4.98% and 14.96% on the NIST 2007 LRE 30-second, 10-second and 3-second evaluation databases, respectively, for the closed-set test condition.

A 2 GHz gain equalizer for analog signal transmission using a novel gain compensation method is described in this paper. This method is based on feedforward compensation by a low-pass filter, which improves the gain-equalizing performance by subtracting low-pass filtered signals from the directly passed signal at the end of a transmission line. The advantage of the proposed method over the conventional one is that the gain is equalized with a smaller THD at higher frequencies by using a low-pass instead of a high-pass filter. In this circuit, the peak gain is adjustable from 0 to 2.4 dB and the frequency of the peak gain can be controlled up to 2 GHz by varying the value of an external capacitor. Also this circuit achieves THD with 5 dB better than the conventional circuits.

Blood pressure is the measurement of the force exerted by blood against the walls of the arteries. Hypertension is a major risk factor of cardiovascular diseases. The systolic and diastolic blood pressures obtained from the oscillometric method could carry clues about hypertension. However, blood pressure is influenced by individual traits such as physiology, the geometry of the heart, body figure, gender and age. Therefore, consideration of individual traits is a requisite for reliable hypertension monitoring. The oscillation waveforms extracted from the cuff pressure reflect individual traits in terms of oscillation patterns that vary in size and amplitude over time. Thus, uniform features for individual traits from the oscillation patterns were extracted, and they were applied to evaluate systolic and diastolic blood pressures using two feedforward neural networks. The measurements of systolic and diastolic blood pressures from two neural networks were compared with the average values of systolic and diastolic blood pressures obtained by two nurses using the auscultatory method. The recognition performance was based on the difference between the blood pressures measured by the auscultation method and the proposed method with two neural networks. The recognition performance for systolic blood pressure was found to be 98.2% for 20 mmHg, 93.5% for 15 mmHg, and 82.3% for 10 mmHg, based on maximum negative amplitude. The recognition performance for diastolic blood pressure was found to be 100% for 20 mmHg, 98.8% for 15 mmHg, and 88.2% for 10 mmHg based on maximum positive amplitude. In our results, systolic blood pressure showed more fluctuation than diastolic blood pressure in terms of individual traits, and subjects with prehypertension or hypertension (systolic blood pressure) showed a stronger steep-slope pattern in 1/3 section of the feature windows than normal subjects. The other side, subjects with prehypertension or hypertension (diastolic blood pressure) showed a steep-slope pattern in front of the feature windows (2/3 section) than normal subjects. This paper presented a novel blood pressure measurement system that can monitor hypertension using personalized traits. Our study can serve as a foundation for reliable hypertension diagnosis and management based on consideration of individual traits.

This work presents a two-stage ordinal optimization theory-based approach for solving the throughput maximization problems with power constraints of sub-carrier assignment and power allocation in multi-user orthogonal frequency division multiplexing uplink systems. In the first stage, a crude but efficient model is employed to evaluate the performance of a sub-carrier assignment pattern and the genetic algorithm is used to search through the huge solution space. In the second stage, an exact model is employed to evaluate s best sub-carrier assignment patterns obtained in stage 1 and form the select subset. Finally, the best one of the select subset is the good enough solution that we seek. Via numerous tests, this work demonstrates the efficiency of the proposed algorithm and compares it with those of other heuristic methods.

To further suppress the multiple-access interference (MAI) in frequency-hopping multiple-access (FHMA) system, a novel kind of FH sequence set named as no-hit-zone (NHZ) sequence, is proposed for an FHMA system with M-ary FSK modulation (MFSK/FHMA) in this paper. Expressions for the decision variables are derived for the asynchronous MFSK/FHMA system with NHZ sequence set (MFSK/FHMA-NHZ) under a slow Rayleigh-fading channel model. For the special case of M=2, accurate analytic bit error rate (BER) is derived as a function of maximum relative delay D and the number of users K by a characteristic function method. The theoretical results validated by Monte Carlo simulations are used to investigate the dependence of the average BER on D and K. Comparison with the MFSK/FHMA system with Markov hit pattern (MFSK/FHMA-Markov) shows that MFSK/FHMA-NHZ system performs better than Markov hit pattern system as long as D is restricted in a certain range and further, the gain in the performance increases with increase in the value of K.

Some new generalized cyclotomic sequences defined by C. Ding and T. Helleseth are proven to exhibit a number of good randomness properties. In this paper, we determine the defining pairs of these sequences of length pm (p prime, m ≥ 2) with order two, then from which we obtain their trace representation. Thus their linear complexity can be derived using Key's method.

In this letter, we analyze the effect of the size of observed data on the performance of time delay estimation (TDE) in the chirp spread spectrum (CSS) system. By adjusting the size of observed data, we reduce the effect of DC offsets, which would otherwise degrade the performance of TDE based on CSS, and we optimize the performance of TDE in CSS system. Finally, we derive the optimal size of observed data of TDE in CSS system.

The H.264/AVC standard achieves significantly high coding efficiency if multiple block size Motion Estimation is adopted. However, the complexity of Motion Estimation and DCT is dramatically increased as a result. In previous work we propose an early mode decision algorithm to control the complexity, based on all-zero-blocks detection in 1616 size. In this paper, we improve the algorithm. Firstly, we propose to detect all-zero blocks in 1616, 88 and 44 sizes to simplify the course of mode decision. Secondly, we define the thresholds which are used to terminate motion estimation and mode decision in advance for these sizes. Last, we present the whole proposed algorithm. Experiments show that about 77% encoding time and 85% motion estimation time can be saved on average, which is better than state-of-the-art approaches.

Over the recent years, a great deal of effort has been made to estimate age from face images. It has been reported that age can be accurately estimated under controlled environment such as frontal faces, no expression, and static lighting conditions. However, it is not straightforward to achieve the same accuracy level in a real-world environment due to considerable variations in camera settings, facial poses, and illumination conditions. In this paper, we apply a recently proposed machine learning technique called covariate shift adaptation to alleviating lighting condition change between laboratory and practical environment. Through real-world age estimation experiments, we demonstrate the usefulness of our proposed method.

To meet the increasing demand to expand wavelength division multiplexing (WDM) transmission capacity, ultrahigh spectral density coherent optical transmission employing multi-level modulation formats has attracted a lot of attention. In particular, ultrahigh multi-level quadrature amplitude modulation (QAM) has an enormous advantage as regards expanding the spectral efficiency to 10 bit/s/Hz and even approaching the Shannon limit. We describe fundamental technologies for ultrahigh spectral density coherent QAM transmission and present experimental results on polarization-multiplexed 256 QAM coherent optical transmission using heterodyne and homodyne detection with a frequency-stabilized laser and an optical phase-locked loop technique. In this experiment, Raman amplifiers are newly adopted to decrease the signal power, which can reduce the fiber nonlinearity. As a result, the power penalty was reduced from 5.3 to 2.0 dB. A 64 Gbit/s data signal is successfully transmitted over 160 km with an optical bandwidth of 5.4 GHz.

A differential cross-correlation cell ID identification algorithm is proposed for IEEE 802.16e OFDMA cellular system. The cell ID represents the number of the preamble selected by the base station in downlink mode. First, we construct the downlink (DL) preamble structure and signal model with carrier frequency offset (CFO) and channel effects. Next, in order to achieve the initial synchronization, a differential receiver with cross correlation for all preamble patterns is proposed to search for cell ID. Simulation results confirm that the proposed structure is suitable for ITU fading channels and outperforms the conventional cell search system.