Applied Mechanics and Materials Vols. 373-375

Abstract: This paper proposes a new Average Kullback-Leibler distance to make an optimal feature selection algorithm for the matching score fusion of speaker verification. The advantage of this novel distance is to overcome the shortcoming of the asymmetry of conventional Kullback-Leibler distance, which can ensure the accuracy and robustness of the computation of the information content between matching scores of two acoustic features. From the experimental results by a variety of fusion schemes, it is found that the matching score fusion between MFCC and residual phase gains most information content. It indicates this scheme can yield an excellent performance.

Abstract: We proposed a new application of image processing in this article. Combined with image segmentation, reducing and effective localization algorithm to recognize the user s input, show the key value on the screen.

Abstract: Unmanned Aerial Vehicle (UAV) remote sensing technology has the real-time ability to obtain ground images at earthquake landslide area. It makes it possible to extract landslide disaster information, support emergency rescue and decrease losses at the first second after earthquake. However, UAV remote sensing is prone to produce images with big-slant-angle. Although it is not fit for object measuring by traditional photogrammetry, it is benefit for observing slope at a better angle and extracting slope characteristics quantitatively. In this research, we define UAV big-slant-angle data as side-look images for earthquake landslide, which is based on an UAV close range photogrammetry idea. Image recognition of earthquake landslide area and three dimensional measuring of slope features will be researched. First of all, a projection datum plane along slope normal will be simulated according to the terrain surface. Appropriate images will be selected by evaluating camera pose and landslide overlapping rate. Secondly, orthogonal rotation matrixes are used to do 3D datum transformation with big-rotation-angle. Settlement of collinearity equations considering big-slant-angle condition will be researched. On this basis, dependent relative orientation among image series will be done to construct a three dimensional side-look-image model. At last, landslide image recognition, slope terrain surface reconstruction and landslide feature measurement is considered to cover the contents of earthquake landslide information extraction. Structural features and combined features can be analyzed and calculated on this basis. A rigorous solution of forward intersection in close range photogrammetry with big-slant-angle will be researched. Slope feature geometry and its relationship among images are to be considered to make a credible characteristic matching thinking big-rotation-angle exists. It has potentiality to provide variety landslide information with higher geometrical precision for cartography and risk analysis of earthquake landslide.

Abstract: Attacks on text-based CAPTCHAs are getting more sophisticated. Most existing schemes defend attacks by increasing deformation or distortion rate of words which sacrifice the friendliness of the system. In this paper, we propose a secure one time text-based CAPTCHA scheme which can effectively defend multiple attacks without decreasing the friendliness of the scheme for valid users. The core of our design is an information obfuscation scheme which determines which character image to display and which to hide in an online manner from a sequence of 62 character images which makes it more difficult to attack. Encryption method is employed for protecting display policies which control showing and hiding of information. Prototype has been implemented and the preliminary results are encouraging.

Abstract: A real-time image data acquisition and processing system based on SOPC in intelligent cart with automatic tracking function was designed. In this paper, the basic design theory and the system structure was analyzed. The Quartus II and Nios II software was used to make the system with real-time image acquisition function. And the data in SDRAM of two CMOS cameras based on SOPC was read. The design was tested with the system, and results showed that this module had the characters of design flexibility, fast image processing and expansibility.

Abstract: The facial expression recognition technology has been widespread concerned and researched, and many methods have been presented. This paper focuses on studying and analyzing the feature extraction, feature dimension reduction and two-against-two multi-class Support Vector Machine (SVM) method, and an algorithm is proposed for recognition of six basic facial expressions. According to expression feature information in the different face region, the algorithm adopts local nonuniform feature point extraction to reduce the feature dimension. After transforming the feature points with Gabor filters, the Gabor expression features are obtained. And the feature dimension is further reduced by discrete wavelet transform (DWT) and discrete cosine transform (DCT). At last, the tow-against-two classification method and an optimum decision scheme are used to realize quick and accurate expression classification. Experimental results show the algorithm can achieve higher recognition rate, recognition speed and stronger robust.

Abstract: A novel signal denoising method using Sym3 wavelet in FMCW Radar lever measurement is proposed. The method provided the signal energy distribution display with respect to the particular time and frequency information. Firstly, the main component of echo signal is extracted by energy analysis and scale decomposition by Sym3 wavelet. Then, the disturbed component is eliminated by frequency filtering. Finally, spectral estimation method is used to estimate the relative distance of the object. The simulation results have shown that the Sym3 wavelet was found to be a better approach for denoising in FMCW Radar lever measurement. The Emulation results have shown that measuring accuracy can achieved the 20 millimeter level which can satisfy the desire requirements of the measurement system.

Abstract: Ventricular Fibrillation (VF) and Ventricular Tachycardia (VT) are arrhythmia which seriously endangers the patient's life, as we know that electrical defibrillation is a most effective way that can rescue their lives. There are some similarities among signals of ventricular flutter (VFL), cardiac arrest, VF and VT; so erroneous judgments may occur by shock able (ShR) signal detection algorithm, which made the patients suffer unnecessary electric shocks. For this reason, a novel method which can distinguish VF and VT rapidly and accurately by ShR signal detection is urgently needed. Approximate entropy, which is a nonlinear dynamic parameters used for measuring sequence complexity and statistical quantification, has been attempted to distinguish various arrhythmias, and preferable results have achieved.

Abstract: In everyday life, people will inevitably cause tension, excitement and normal emotional state, emotional state, is very easy for any crowd collected EEG analysis, but the crowd, so nervous, excited state generated for different reasons, it is difficult to acquisition-related EEG characteristics were analyzed. College students are a special population, caused to them by the state of tension and excitement similar analysis, were college students before and after the final exam results announced before the announcement of the scholarship and the announcement of the scholarship, as well as state EEG acquisition and analysis to analyze EEG characteristics under tension, excitement and normal state of college students.

Abstract: Based on the characteristics of the acoustic emission (AE) signals from low carbon steel pitting corrosion, a new extraction method was proposed with wavelet transformation and independent component analysis. The experiment result shows that the new method can overcome the influence induced by the uncertainty of the independent source of low carbon steel pitting corrosion and good extraction result can be achieved.