Applied Mechanics and Materials Vols. 278-280

Abstract: Fuzzy linear regression has been extensively studied since its inception symbolized by the work of Tanaka et al. in 1982. As one of the main estimation methods, fuzzy least squares approach is appealing because it corresponds, to some extent, to the well known statistical regression analysis. In this article, a restricted least squares method is proposed to fit fuzzy linear models with crisp inputs and symmetric fuzzy output. The paper puts forward a kind of fuzzy linear regression model based on structured element, This model has precise input data and fuzzy output data, Gives the regression coefficient and the fuzzy degree function determination method by using the least square method, studies the imitation degree question between the observed value and the forecast value.

Abstract: Anti-aliasing algorithm for circle was complex and the effect was not satisfactory. Based on midpoint generation algorithm for circle, an integral algorithm was present for circle anti-aliasing. The algorithm assigned the grayscale of each pixel according to the distance between a pixel's center and the circle, so a circle with various grayscales can be drawn. The algorithm abandoned the two-order epsilon and corrected the error by simple calculation. The algorithm was explained using 6 bits grayscale anti-aliasing as an example based on researching on the relation between precision and complication. The algorithm can forecast the grayscale change between neighboring pixels using integer without floating-point, so it is convenient to realize on hardware. The result shows that the anti-aliasing effect was improved.

Abstract: Due to the disadvantages of genetic algorithm such as the weaker ability for local search, premature convergence, random walk and problems related, and so on , the design and improvement of the algorithm is an important research direction of genetic algorithm. And evaluating the performance of algorithm systematically and scientifically is the key to test algorithm whether good or bad .The common method used to evaluate algorithm is test function, however, the existing literature on the optimization algorithm has different methods to evaluate the performance of algorithm, and there is no uniform test criteria. As for those questions above, This paper studies test functions of genetic algorithm, and analyses characteristics of the main test functions, which can be used as the basis of selection algorithm test functions.

Abstract: It is difficult to evaluate water quality, because there are lots of influence factors. It presents discrete Hopfield neural network to evaluate water quality in this paper. The criteria of nutrition water quality used as training samples to train discrete Hopfield neural network, which is stored in discrete Hopfield neural network. The data in a monitoring point used as test samples to evaluate water quality. The experimental results show that discrete Hopfield neural network can evaluate water quality effectively. The classification results can be shown directly, and the running speed is faster than BP neural network. This method has important guiding significance and practical application value to other water quality evaluation.

Abstract: This paper suggests an improved genetic algorithm to seek the minimum range value in the ideal-plane flatness measurement. This algorithm increases measurement accuracy by using dynamic cross factor, mutation factor and a new concept called chromosome fitness. It was proved in simulation experiments that its accuracy is better than other flatness error evaluating algorithms like the minimal territory evaluating algorithm and the computational geometry algorithm etc. So it can be used for measuring industrial production components error and verifying assumed models in reverse engineering etc.

Abstract: A novel method of soft sensing is propsed combined Kernel Isomap (KIsomap) with Least squares support vector machines (LS-SVM). KIsomap is an improved Isomap and has a generalization property by utilizing kernel trick. It is a kind of novelly promoted nonlinear methods for dimension reduction, and can effectively find out the intrinsic low dimensional structure from high dimensional data. The KIsomap is used to feature extraction and reduce dimensions of sample. The LSSVM is applied to proceed regression modelling, which can not only reduce the complexity of modeling but also improve the generalization ability.The proposed method is used to build soft sensing of diesel oil solidifying point. Compared with other two models, the result shows that KIsomap-LSSVM approach is effective and correct.

Abstract: In this paper, we study the neural networks with time delays. By using of the Homeomorphism theory and employing an inequality, constructing a new Lyapunov-Krasovskill functional, we give a new sufficient condition, which is independent of the delays, guarantying the existence, uniqueness of the equilibrium point and it’s global exponential stable. Meanwhile, the incorrectness of calculation in the previous Refs is pointed out. We therefore improve the previous results.

Abstract: In order to reduce Gaussian and Salt & Pepper noises, a combination approach to noise reduction is presented by combining the median filter with the mean filter. The detail simulations show that the mode which the median filtering first and then the mean filtering is superior to that of the simply single filtering, or the mean filtering first and then the median filtering when the image obviously contain the Salt & Pepper noise. On the other hand, it is not necessarily the optimal scheme to use the mode which the mean filtering first and then the median filtering when the digital image obviously contains the Gaussian noise.

Abstract: The copyright protection of multimedia information is more and more important in the digital age, digital watermarking is a solution to address the topic. This paper proposes a new robust image watermarking scheme based on discrete ridgelet transform (DRT) and discrete wavelet transform (DWT). The scheme respectively embeds a copy of color image watermark into DWT domain and DRT domain, and uses a conception of semi-watermark. Experiment results demonstrate that the watermark can resist various attacks such as adding Gaussian or Union Distribution Noise, JPEG compression, brightness adjustment, contrast adjustment, altering color balance, lens blur, zooming in or out, cropping and some combined attacks etc.

Abstract: Abstract. In order to figure out the low-cost imaging apparatus’ problems, such as the low image contrast, the indistinct vein characteristics, which are affected by the photoelectric noise during the collection process. In this paper, the vein imaging principle is discussed, and a low-cost collection device is designed, at the same time a two-dimensional multi-directional Gaussian filter is constructed to achieve the enhancement of low quality palm vein image, and the Local Dynamic Threshold Segmentation Method (Niblack) is adopted to extract vein characteristics. The results show that this method is able to extract distinct vein characteristics, and also can provide reliable characteristics basis for palm vein recognition system.