Applied Mechanics and Materials Vols. 644-650

Abstract: The terahertz (THz) spectrum technique has become an important research content of electromagnetic field in recent years. Folded waveguide backward wave oscillator (BWO) is suitable to be used as THz high-power electromagnetic radiation source. Because it can realize electronic tuning within a wide frequency band and doesn’t require pre-stage source amplifier. In this paper, the high frequency characteristic, the transmission characteristics, and the particle beam wave interaction characteristics of the structure of folded waveguide BWO have been analyzed, and it is utilized to design geometry parameters of 650GHz folded waveguide slow-wave-structure (SWS). The design is been validated by particle beam wave interaction simulation experiments. The experiment results show that the output is over 1.7W within 42GHz bandwidth, and the electron efficiency can exceed 1%.

Abstract: LMS algorithm is a kind of classic adaptive algorithms. Although it has the virtue of simple operation, it also shows the defects of relatively slow convergence and big steady state errors in low SNR. To remedy these defects, this paper put forward a new variable steps adaptive LMS algorithm. In the transient state, the learning rate increases slowly with the iteration times which accelerate the convergence rate of LMS algorithm. In the steady state, the learning rate decreases gradually with the iteration times which guarantee the convergence accuracy of LMS algorithm. After this improved algorithm is applied in the design of adaptive wavetrap, the simulation results show that it can not only effectively ease up the conflicts between convergence rates and steady state errors, but also improve the performance of wavetrap in real-time trapping.

Abstract: As an important property of the quantum dot infrared photodetector, the noise has attracted extensive attention. In this paper, the model for the noise of the QDIP is built. This model takes the total electron transport and the dependence of the drift velocity of electrons on the electric field into account. The corresponding calculated results not only show a good agreement with the published results, but also illustrate the dependence of the noise on the structure which can provide us with a method used to optimize the structure of the detector devices.

Abstract: An image denoising method based on spatial filtering is proposed on order to overcoming the shortcomings of traditional denoising methods in this paper. The method combined mean mask algorithm with median filtering technique is able to replace the gray values of noisy image pixel by the mean or median value in its neighborhood mask matrix and highlight the characteristic value of the image. Peak signal to noise ratio and mean square error are used as the evaluation index in this method and comparison between mean filter and median filter is done. The experimental results show that this denoising system makes the images have a high signal to noise ratio and integrity of edge details and take into account real-time, and fast response characteristic of the system.

Abstract: The full-waveform technology in the small footprint airborne LIDAR systems has enabled us to sample and record the whole returned waveform of a laser shot. Under most conditions the full-waveform signals satisfy the K-sparse condition and thus can be sampled and recovered using the theory of compressive sensing. This paper proposes a new compressive-sensing-based data acquisition and processing method, by which the returned signals are sampled in small number by a Toeplitz-like matrix, and the target cross sections and the full-waveforms are approximated and recovered by GPRS algorithm. The feasibility of this method is verified by simulated experiments under both ideal conditions and non-ideal conditions with noises. The advantage of this method lies in that requirement for the slew rate of the ADC is lowered, which is convenient for the data storage and transmission .

Abstract: This paper mainly studies modeling and recognition of 3D English words’ images. With the development of secondary modeling, segmentation and recognition theories and the application of evolution computation in 3D modeling and recognition, this paper analyzes the issues of parameter fitting in the 3D model, multi-object scene segmentation and parts recognition aiming at the 3D data features in the English words. The 3D model is used as the primitives part to model and segment the scenes and the group parallel evolution and the relationship matching theories are introduced into the 3D modeling and recognition to deeply identify the rare English words’ images. The paper searches for a practical and efficient three-dimensional modeling and identification scheme.

Abstract: This paper mainly studies the recognition issue of blurred alphabet images. If alphabets in an image is blurred, it is difficult t be recognized. To avoid the defect, the paper proposes a computer recognition method for blurred alphabet images based on grayscale class variance algorithm. The algorithm executes non-linear transformation on feature parameters of alphabets images extracted to obtain eigenvector coefficient weights, and then calculates characteristic correlation coefficient through wavelet transform to realize the blurred alphabet recognition. Experiments show that, the proposed method improves the accuracy of the blurred alphabets recognition and achieves satisfactory results.

Abstract: In the technology field of world's high-speed Computer Numerical Control (CNC) communications, accelerating research for remote communications signal technology will not only promote the development of China's high-speed remote technology, but also has far-reaching scientific value, industrial and economic value to the train remote communications. The new train remote communication technology is the basic topic with high technology, which has complicated procedures, in fact, contains mixture and summary of multi-skills, such as the compute control, application of sensor technology, the recognition of test skills and digital signal processing, and aggregation. Therefore, it can improve signal collection efficiency, and access to dynamic unbalance signal of the train remote communication signal.

Abstract: The locomotive frequency shift signals carry important operation information. In order to achieve reliable detection of the locomotive signals, this paper analyzes the signal characteristics of the locomotive and proposes a railway remote communication signals collection model. The model uses the data mining methods to extract the locomotive frequency shift signals by analyzing the locomotive frequency shift with signal classification technology. The separated signals can obtain the low frequency signals with low-pass filtering and shaping. The algorithm is simulated by MATLAB software. The results illustrate the proposed method can effectively separate and collect the shifted frequency signals with some application value.

Abstract: The paper proposes a kind of athletes’ irrational postures recognition technology based on minimizing image sequences by analyzing the shape bases of the collected images and estimating the overall movement parameters in the single-frame movement images. The irrational movement features information will be projected in the gray fields to obtain the embedded hidden information in the overall movement posture images to achieve the recognition of irrational postures. The experimental results illustrate the posture correction offsets by the proposed improved algorithm are smaller than those in traditional methods with high correction accuracy and the signal to noise ratios of the peaks are increased. The results show the superiority of the algorithm which can be well applied in the motion image fusion and identification areas.