Abstract:In this paper, the principle of fiber laser hydrophone was presented, and the development of fiber laser hydrophone was described. Researches on fiber laser hydrophone technology of our group were reviewed. It contains technologies on fiber laser hydrophone, fiber laser vector hydrophone, hydrophone array, field tests and application areas. The static pressure sensitivity of fiber laser hydrophone is better than－140 dB re 1 pm/μPa, from 20 Hz to 2000 Hz. The diameter of towed array is thinner than 30 mm, and the number of its elements is up to 64. Fiber laser hydrophone technology is coming into application from laboratory. We believe that the researches on practical application are of great importance.

Abstract:The phytoplankton groups were investigated in 30 shallow Chinese lakes in the middle and lower reaches of the Yangtze River area in the summer of 2003 and 2004. To explore the effects of different nitrogen phosphorus ratios (TN/TP ratios) on the phytoplankton taxonomic composition of these study sites, six main taxonomic groups were studied with three TN/TP ratios intervals: TN/TP＞30, 12＜TN/TP＜30 and TN/TP＜12. The biomass curves of these taxonomic groups showed corresponding increases or decreases with different TN/TP ratios. When TN/TP ratios declined from high (＞30) to medium (12-30), the slopes of the total biomass curve increased, as did the relative abundances of all groups except Cyanophyta. But when the TN/TP declined from medium (12-30) to low (＜12) levels, the slopes of most groups decreased except Chlorophyta and Cryptophyta. The amount of Cyanophyta increased with TP when TN/TP ratios declined from above 30 to below 12, suggesting that cyanobacteria adapted to higher TN/TP ratios. However, Chlorophyta and Cryptophyta tended to be restricted by phosphorus when TN/TP ratios declined from above 30 to below 12, and these groups adapted to lower TN/TP ratios. The biomass of Bacillariophyta, Pyrrophyta and Euglenophyta tended decreased when TN/TP ratios were above 30 and below 12, suggesting that medium TN/TP ratios (12-30) favoured these groups.

Abstract:The Mandarin Chinese language is known to be strongly influenced by a rich set of regional accents, while Mandarin speech with each accent is of quite low resource. Hence, an important task in Mandarin speech recognition is to appropriately model the acoustic variabilities imposed by accents. In this paper, an investigation of implicit and explicit use of accent information on a range of deep neural network (DNN) based acoustic modeling techniques was conducted. Meanwhile, approaches of multi-accent modeling including multi-style training, multi-accent decision tree state tying, DNN tandem and multi-level adaptive network (MLAN) tandem hidden Markov model (HMM) modelling were combined and compared. On a low resource accented Mandarin speech recognition task consisting of four regional accents, an improved MLAN tandem HMM systems explicitly leveraging the accent information was proposed, and significantly outperformed the baseline accent independent DNN tandem systems by 0.8%-1.5% absolute (6%-9% relative) in character error rate after sequence level discriminative training and adaptation.

Abstract:Automatic registration of point clouds is a challenging task especially when the overlap between them is too small to initialize the traditional iterative closest point algorithm directly. A method for registering 3D point clouds in different coordinates was proposed by using the scanner’s pose information, recorded by a robot. This method consisted of two steps: firstly, 3D scanner was set on the end effector of the robot, which recorded the 6D pose of the scanner when an object was scanned in real time. Using this recorded pose information, the captured point clouds from different scanner coordinates were transformed to robot base coordinate. Secondly, the weighted sparse iterative closest point was used to align the point clouds in robot base coordinate which refines the result of the first step. This method was tested on various data and situations.The experiment results show that the proposed method could align point clouds with lower overlapping ratio, and is more accurate, faster and more robust to outliers than existing methods.

Abstract:In order to extract components from the target system effectively and achieve the goal of software reuse, a component extraction method based on design documents was proposed in this paper. Computer aided software engineering tools were used to design the target system, then concept components and system architecture were extracted from this design documents. After the accuracy of extract result was verified by extensible markup language element and matching degree of characteristic element, foundation for software reuse was provided in this way. The application of this method was illustrated through two practical applications: product resource management platform and change-major student system. During this process, concept components of fileUpload and dataDump were extracted and verified, and components which match these concept components were reused successfully for the system. The result shows that the proposed method not only provides a framework of reusable components, but also improves the success rate of software reuse.

Abstract:The Internet of Things realizes the connection of human and objects. Activity recognition is necessary for the interaction between information sensing devices and human. Currently, vision-based and sensor based methods are widely used, but these methods are limited in many scenes. In this paper, a new radio-frequency-based activity recognition technique was proposed, in which a few communication nodes were deployed in the monitoring area for the device-free activity recognition by analyzing the transmission packet state information. The sequential minimal optimization and K-nearest neighbor algorithms were employed for classification. The classification accuracy of walking speed of the proposed method is improved by 25.1% on average compared to the traditional method based on received signal strength indication.

Abstract:Optogenetics has been successfully applied to understand the mechanisms of neuropsychiatric diseases through the precise temporal control of specific neural circuitries. However, it remains a great challenge to integrate optogenetic modulation with electrophysiological recordings in multiple brain regions in vivo. In this study, a simplified method for the fabrication and electrochemical modification of the multicircuit optrode arrays was developed. The modified optrode arrays exhibited a significantly higher capacitance and lower electrochemical impedance at 1 kHz as compared to unmodified optrodes. The optrode arrays were chronically implanted into the brain of VGAT-ChR2 transgenic mice. Spontaneous action potentials and local field potentials as well as light-evoked responses were obtained in 4 different brain regions in vivo. The crossarea synchronizations were analyzed and the localizations of the implanted optrode arrays were confirmed by 4', 6-diamidino-2-phenylindole immunofluorescence staining. All these characteristics are greatly desired in optogenetic applications, and the fabrication method of the optrodes can be easily integrated with other in vivo techniques to build more advanced tools for the dissection of neural circuitry.

Abstract:Virtual navigation in 3D scenes requires both smooth and intelligent camera control, which is a very challenging task. In this paper, a real time navigation system that could automatically generate an effective and smooth camera flyby trajectory was presented. Firstly, offline geometric and semantic features of models were analyzed in the given 3D scene, including buildings, roads and other visible parts, to evaluate the importance of models. Secondly, the path from origin to destination was sampled adaptively, and visual images from different views on each sampled point were stored effectively. Finally, the best view on each sampled point was extracted and connected smoothly to get the camera flyby trajectory. The experiments done in four different 3D scenes clearly demonstrate the efficiency and intelligence of the proposed algorithm, and the user investigation also verifies its superiority over the existing methods.

Abstract:Since time constant and currents could be affected by inductance and resistance after coil combined transformation, a method of coils segmentation and transformation was used for the current-carrying coils. The response time of voltage loading, the output electromagnetic force and the acceleration of an electromechanical converter (EMC) were studied, respectively. Compared with single coils in series, the frequency of EMC with multiple coils in parallel could be about 500 Hz, both the length of coils and current are increased, the peak of output electromagnetic force of EMC is 276.4 N, which is over 16 times larger than that of single coils in series, and the acceleration is up to 27.64 g. Analysis results show that the proposed methods display the characteristics of high response and large thrust force.