Applied Mechanics and Materials Vol. 740

Abstract: An improved image segmentation algorithm based on watershed transform is presented In this paper. Firstly the normalized cut method and watershed transform are explained and analyzed. Secondly the idea of the improved algorithm and the main formula are explained. We consider the area and perimeter when we merge adjacent regions. We define a new weight value and discuss the value of the parameter α and β. Finally the experiment result is presented. The new algorithm reduces the nodes and the computational demand of the common normalized cut technique.

Abstract: Electronic image stabilization (EIS) is the technology which detects and removes camera jitter in image sequences by digital image processing methods. It has been widely used in moving vehicles such as airborne, shipboard and vehicle-mounted camera systems. This paper firstly introduces the basic principle and structure of EIS system. Then the key techniques for each module are introduced and compared in processing speed and application scenes. The motion estimation method based on feature points is emphatically analyzed. Some experimental simulations and further discussion are made on the existing motion filtering and compensation methods. We also summarize the evaluation criteria for EIS performance. Finally the future research trend of EIS system is presented.

Abstract: The approximate merging of two adjacent B-spline surfaces into a B-spline surface is the core problem in data communication. A novel algorithm is presented in this paper to solve this problem. In this algorithm, we compute the merging error using L₂ norm instead of the Euclidean norm, the process of merging is minimizing the approximate error and we only need solve a system of linear equations to get the final merging surface. In order to reduce the merging error, we add a weighed function on objective function to start the next merger; this function adds greater weight on where error is larger. After the next merger, the merging error will be significantly reduced. Finally, some examples are given to demonstrate the effectiveness and validity of the proposed algorithm.

Abstract: K-means is one of the most widely used algorithms for clustering. Ease of implementation, efficiency, simplicity, and empirical success are the main reasons for its popularity. In actual application, there are some defects in traditional k-means, for example, the value of K need to be specified ahead, initial clustering center is a random choice and so on; this influences the performance of the K-means. In order to overcome these obstacles, many variants of K-means algorithm have appeared. We provide a brief overview of k-means, point out existing problems; summarize major improvements in the determination of clusters number, the initialization of the cluster, the similarity measurement, the sensitivity of noise and outliers and so on. Further study directions of K-means are pointed at last.

Abstract: In this paper, we present an algorithm for surface approximation to the measured cloudy data with four boundaries. Several key techniques about this algorithm are also described, such as base surface construction, projection and weighted least square approximation. We use weighted least square to reduce the times of iterations, because the iteration is a very expensive and error prone process. We add a positive weight to each point, and the weight-adding algorithm is introduced as well. Increasing the weight onto this point will decrease the approximation error of the point. Finally, some examples of this algorithm demonstrate its effectiveness and validity.

Abstract: We present an automatic method to identify cataclysmic variable stars (CVs) in the ninth data release of the Sloan digital sky survey (SDSS). The data mining technique is employed and the massive spectra are identified quickly and efficiently. The high dimensional spectra are mapped to feature space constructed by the principal component analysis (PCA), and dimensionality reduction is carried out accordingly. Massive SDSS spectra are classified by the support vector machine (SVM) and most of the non-candidates are excluded. The final greatly reduced candidates can be identified manually and easily. Experiments show that this data mining method can find CVs from the SDSS database in an effective and efficient manner. In addition, this method is also applicable to mining other special celestial objects in sky survey telescope data. We report the identification of a new CV with spectra. The newly found CV enriches the CVs spectral library and will be useful to the research of binary evolution models.

Abstract: Singular Value Decomposition used in spectrum feature extraction, often discards small component that may be important for identifying mineral oil products. This work presents a new method using the Singular Value Division (SVD) on Wavelet Transform (WT) with three-dimensional fluorescence spectra as the source of oil features. WT-SVD feature based fuzzy classification (FCM) is implemented and comparable or better results are yielded in more accurate, and more robust than SVD performance under random noise conditions. The result means that WT-SVD method can strike a balance between data compression and preservation of small valid information in feature extraction of three-dimensional fluorescence spectra of mineral oils. This method is conducive to oil discrimination and pollution analysis in water environment monitoring.

Abstract: Wavelet transform denoising is an important application of wavelet analysis in signal and image processing. Several popular wavelet denoising methods are introduced including the Mallat forced denoising, the wavelet transform modulus maxima method and the nonlinear wavelet threshold denoising method. Their advantages and disadvantages are compared, which may be helpful in selecting the wavelet denoising methods. At the same time, several improvement methods are offered.

Abstract: Internet of Things is envisioned as promising technologies for remote equipment maintenance in large marine ships. Data aggregation is critical for sensing data collection in Internet of Things. The network design affects data aggregation efficiency. As the volume of sensing data is large due to the number of equipments, it is mandatory to decrease the communication overhead in data aggregation. In this paper, we propose a weighted graph based network design algorithm for data aggregation, called Endada. The communication efficiency is improved by Endada, which is justified extensively by formal analysis and rigorous proof.

Abstract: Interlaced scanning has been widely used as a trade-off solution between picture quality and transmission bandwidth since the invention of television. During the past decades, various interlaced-to-progressive conversion algorithms have been proposed to improve subjective quality or coding efficiency. However, almost all the researchers concentrate on general cases, without making full use of specific application scenarios. Based on extensive investigations, eliminating visual artifacts in areas of subtitles and station captions for interlaced sports and news videos is still an unsolved problem, which will be addressed in this paper. Firstly, motion estimation is performed between field pictures. Secondly, text edge detection is proposed for sports and news videos. Finally, different processing strategies are applied to text regions and non-text regions. Experimental results show that the proposed method can generate much better text content than existing algorithms. In addition, it is quite stable for non-text parts.