Applied Mechanics and Materials Vols. 536-537

Abstract: On the basis of analysis the plate quality detection method in accordance with machine vision, Innovative dot area coverage percentage measurement method for plate microscopic image based on color segmentation was proposed in this paper. Plate microscopic image was transformed from RGB to HSI model. Then the plate microscopic image was segmented according to the saturation and the hue information. It was proved by experiments that the dot area coverage percentage measurement method based on color segmentation was provided with high measurement precision and good plate applicability.

Abstract: Speech synthesis is a hot research of artificial intelligence today, and urgent difficulty to overcome is how to make the machine more "emotional intelligence" for the human-computer interaction. With the STRAIGHT algorithm, this paper extracted the acoustic feature parameters of the speech signals and did statistical analysis, modified the characteristic parameters of the neutral sounds to synthesize emotional speeches, including happy, angry, frustration, then analyzed a frame of spectrum of synthetic emotional speeches through standard voices, voices added noise and voices de-nosing. The experimental results show that the method is feasible and the synthetic emotional speeches through voices de-nosing are better than voices added noise.

Abstract: Curvelet transform is the combination of the multi-scale analysis and multi-directional analysis transforms, which is more suitable for objects with curves. Applications of the curvelet transform have increased rapidly in the field of image fusion. Firstly, using the curvelet transform, several polarization images can be decomposed into low-frequency coefficients and high-frequency coefficients with multi-scales and multi-directions. For the low-frequency coefficients, the average fusion method is used. For the each directional high frequency sub-band coefficients, the larger value of region variance information measurement is used to select the better coefficients for fusion. At last the fused image can be obtained by utilizing inverse transform for fused curvelet coefficients. In the present work an algorithm for image fusion based on the curvelet transform was implemented, analyzed, and compared with a wavelet-based fusion algorithm. Experimental results show that the proposed algorithm works better in preserving the edges and texture information compared with the wavelet-based image fusion algorithms.

Abstract: The Gabor wavelet is the important technique widely used in the areas of images recognition such as human face expression, it extract the more important grain features for face expression effective, but it does not take into account the relative changes in the important characteristics of each location of the point features. Aiming at recognizing the information of human face expression, fuse the geometry feature based on angle changes at key parts on face expression, and then a radial basis function (RBF) neural network is designed as the classifier to perform recognition. The results of the experiment in the human face expression database indicate that the recognition rate by the feature fusion is obviously superior to that of traditional method.

Abstract: In this paper, we analyze the principles of the dark channel prior based on guided filtering image algorithm to defog, pointing out the shortcomings and derive an improved method. Dark channel prior principle is established in the absence of bright areas, which not satisfied Dark channel prior, and thus, the transmittance of the bright areas is estimated error, which will case color distortion of defogged image. By introducing a tolerancemechanism refining the transmittance, the algorithm can effectively handle such problem to overcome the color distortion in bright areas using dark channel prior. Experimental results show that this modification is substantial practicable to restore image, at the same time eliminates the color distortion, significantly improve the visual effect.

Abstract: Based on the shape of the image retrieval occupy an important position in the content-based image retrieval, and studied architecture, content-based image retrieval system, ie research-based image retrieval key technologies shape features for image noise in addition to the morphological processing; image segmentation; shape-based feature extraction and regional boundaries and description techniques and similarity measure techniques. The results show that the algorithm can effectively identify the characteristics of the image.

Abstract: A method is presented to accurate face-mating point cloud registration after dealing with noise point. Point cloud registration is divided into two parts,firstly,coarse registration is applied for visual point cloud,then three local sufaces of overlap cloud region are selected to be mating calculated after denosing base on least squares fitting , at last accurate splicing parameters of translation and rotation are acquired by nonlinear least square .This algorithm is easy to deal with the denosing, has faster convergence speed and higher registration accuracy.Its feasibility is proved by samples.

Abstract: In speech signal processing, the techniques of speech segmentation as front end of preprocessing have great importance in speech enhancing, coding and recognition. This paper analyzes the performances of several typical algorithms of speech segmentation, which are compared with each other. It put emphasis on the study of the algorithm based on the wavelet transformation. The smooth and gradual changing low frequency component can not segment the speech efficiently. In order to solve the problem, this paper put forward to an algorithm based on the cumulate energy of the wavelet transformation which promotes the precision of the segmentation on the phoneme level. But as a result of the wavelet sensitivity, it will present certain number of false spots. Therefore this paper proposes tow methods removing false spots. Finally it makes certain summary to these technologies.

Abstract: Establishing calculation model on the probability of saliency objectness likelihood based on superpixels is introduced to dectect image saliency. At first,analysing factors which affect the size of visual attention according to the law of universal gravitation; And then use the SLIC algorithm to divide image into superpixels; Next, according to the texture, color,and gradient feature information, establish calculation models on probability saliency object under different rules: Including compactness in class color ,spatial distribution estimation and edge continuity; And then refer to the principle that activity in cells responding to stimuli,.a new feature combination theory is proposed to deal with the relationship of the characteristics of independence and mutual interaction for achieving features fusion accurately. Afterwards, the proposed algorithm applied in virtual and reality interaction to detect the effective region and eliminate noise area.

Abstract: This paper researches on the detection and tracking technology of ship intrusion and remaining based on the special surveillance video scenes of sea. A ship detection and tracking method that combines the adaptive background subtraction with centroid algorithm is proposed. Compared with the method of edge orientation histogram, the method proposed in this paper shows a lower computational complexity and a better real-time performance. The experiment demonstrate that this method has a high degree of accuracy. And it makes a realization of detecting and tracking of ship intrusion and remaining within the surveillance sea.