Applied Mechanics and Materials Vols. 571-572

Abstract: In this paper, A method of producing data sample with an inadequate one by combining Monte Carlo method and GM (1,1) Model theory is introduced. This method can produce a larger data sample, which feet the demand of data amount in Neural Networks. This method was examined with reconstructing an inadequate amount data sample of highway vehicle sound. Neural Networks have been found having a good performance in identifying vehicle type with the data reconstructed. Also, the steps to reconstructing an inadequate data sample with Monte Carlo method and GM (1,1) model theory are presented.

Abstract: As One of Features from other Languages, the Chinese Tone Changes of Chinese are Mainly Decided by its Vowels, so the Vowel Variation of Chinese Tone Becomes Important in Speech Recognition Research. the Normal Tone Recognition Ways are Always Based on Fundamental Frequency of Signal, which can Not Keep Integrity of Tone Signal. we Bring Forward to a Mathematical Morphological Processing of Spectrograms for the Tone of Chinese Vowels. Firstly, we will have Pretreatment to Recording Good Tone Signal by Using Cooledit Pro Software, and Converted into Spectrograms; Secondly, we will do Smooth and the Normalized Pretreatment to Spectrograms by Mathematical Morphological Processing; Finally, we get Whole Direction Angle Statistics of Tone Signal by Skeletonization way. the Neural Networks Stimulation Shows that the Speech Emotion Recognition Rate can Reach 92.50%.

Abstract: Disparity estimation is a basic analytic method for images and video and is highly used in fields like stereo image coding, video retrieval and stereo image segmentation. This paper proposes a new disparity estimation method. In this paper, the superiority of the multi-block matching is analyzed. The new method based on the idea of block matching constructs a multi-block matching algorithm, thereby improving matching accuracy especially in the disparity discontinuous regions. A adaptive window method based on edge detect is proposed to select the best window.The experimental results based on the Middlebury stereo datasets demonstrate that our method can achieve a better performance.

Abstract: The streamline is of great importance for the visualization of 3D vector field. In order to solve those problems of streamline occlusion and visual confusion caused by of excessive streamlines. Meanwhile the streamlines have to present the variation law and important features of vector field exactly. In this paper, a view-dependent and physical feature-preservation streamline simplification method for 3D vector field visualization is proposed. Firstly, streamline set of 3D vector field is generated by particle tracking. Secondly, view-dependent mapping of the streamline set is implemented. Finally, the streamline set is simplified effectively following an iterative feature-preservation computation and view-dependent computation. The experiment results show that the streamlines are visually enhanced meanwhile the physical features of vector field are effectively maintained.

Abstract: The objective of this study is to make some improvements to the original Moving Particle Semi-implicit method (MPS) for free surface flows. Compared to traditional mesh methods, MPS is feasible to simulate surface flows with large deformation, however, during the simulation; the pressure oscillation is quite violent, duo to misjudgment of surface particles as well as particles gathering together. To modify this problem, a new arc method is applied to judge free surface particles, and a collision model is introduced to avoid particles from gathering together. Hydrostatic pressure and classical dam break are investigated by original and improved MPS. The results verify that improved MPS method is more effective for free surface flows.

Abstract: For a vessel flexure of measure ship under the influence of internal and external environmental factors, which resulting data errors of measuring equipment, the development of the "Vessel Flexure System" can solve the problem. The paper briefly describes the principle and hardware components of Vessel Flexure System, focuses on the application of digital image processing techniques used on linear CCD, put forward the interpretation method of gravity-center, which can determine the location of the center of facula accurately.

Abstract: Part size measurement is a very common and important projects in the production and life, Image size measurement system is composed of camera and Image processing algorithms, Size measurements is finished on this platform. Pixel equivalent of 0.1198 mm / pixel and measurement range of 0.1198 ~ 42.169 mm is got by calibration experiment, analyzing experimental data, measurement accuracy reaches ± 0.09 mm. Instrument work properly and meet the design requirements in a laboratory environment.

Abstract: As the most common type of facial occlusion, eyeglasses may cause great accuracy degradation in face recognition. In this paper, we proposed an improved approach on automatically detecting and removing eyeglasses from grayscale facial images. First, we normalized a face image by the result of face detection and eyes localization. Then we used a fast judging method to detect eyeglasses’ presence. For images with eyeglasses, we used PCA reconstruction error and edge feature to determine the occluded area, and synthesized the area through image inpainting. Experimental results show that our approach can detect the presence of eyeglasses very accurately and obtain generally natural looking images without eyeglasses. In face recognition test, our approach greatly contributed to the accuracy of recognition, achieving higher improvement than other approaches such as simple PCA reconstruction, iterative error compensation, and weighted fusion.

Abstract: In allusion to the problems that the traditional bilateral filtering algorithm is based on pixel level, the intensity similarity weight values may be disturbed by noises. a modified bilateral filtering algorithm is proposed, which uses wiener function to estimate the values of the centre pixels according to neighbor pixels, thus, noise interference on the weighted coefficient can be reduced. Furthermore, Gauss function is replaced by Geman-McClure (GM) function to improve denoising performance. The experimental results show that the proposed algorithm can get denoised image with higher subjective visual quality and objective evaluation index.

Abstract: Lofting is widely used to approximate the scanned data in row-wise fashion, but this method is prone to result an astonishing number of control points in the process of making the rows curve compatible. A novel algorithm of B-spline surface approximation to the scanned data is presented in this paper to solve this problem. Firstly, the scanned data are interpolated by rows of curves; then these curves are approximated by other curves using least square approximation. In this process, all curves are approximated by a common knot vector, and it is different form the traditional method that each curve is approximated by a different knot vector, so we needn’t insert many knots in each curve to make curves compatible. We also can meet high accuracy without losing the shape of lofting surface because we firstly interpolate the data, the best least square approximation substitute insertion of knots in lofting. Numerical example shows that the proposed method is efficient in reducing control points of the lofting surface.