Search

Abstract: The detection system for frame embossing surface is discussed in this paper. According to the periodic feature of frame embossing surface, one periodic embossing is inspected once. Then a fast template matching algorithm based on projection and normalized cross correlation measure is proposed in this paper. The experimental results showed that the proposed algorithm was more efficient and faster than conventional image template matching algorithms. [Copyright &y& Elsevier]

Abstract: In this paper, we propose an ear recognition system based on 2D ear images which includes three stages: ear enrollment, feature extraction and ear recognition. Ear enrollment includes ear detection and ear normalization. The ear detection approach based on Gentle Adaboost algorithm detects the ear part under complex background using two steps: off-line cascaded classifier training and on-line ear detection. Then Active Shape Model is applied to segment the ear part and normalize all the ear images to the same size. For its eminent characteristics in spatial local feature exaction and orientation selection, Gabor-based ear is presented in this paper. Kernel Fisher Discriminant Analysis is applied for dimension reduction of the high-dimensional Gabor features. Then distance based classifier is applied for ear recognition. The experimental results on USTB ear database show effectiveness of the proposed approach. [Copyright &y& Elsevier]

Abstract: The natural scenes of traffic signs accurately identify the classification of intelligence is an important part of the car. The quantity of traffic signal is big. The shape is complex. And due to the light changing, fading, being distorted on the surface of traffic signs, in natural scene automatic identification is a big challenge. In this paper, we propose a method combined the improved scale invariant feature transform(SIFT) algorithm which reduces the high dimension and high complexity with Affine Scale Invariant Feature Transformation(ASIFT) algorithm under the natural scene traffic sign detection and recognition. Experiments have proved that the method has a high recognition rate and the rapid pace of identification, intelligent vehicle system with a higher value. [Copyright &y& Elsevier]

Abstract: The timeliness of system is restricted by processing speed of images in aerial photogrammetry. This paper focuses on the feature points extraction and matching of images based on unmanned airship. A simplified algorithm, which increases the speed of extraction through pixel-spaced sampling and the speed of matching through reducing the search band, is introduced. RANSAC algorithm is used to estimate fundamental matrix. In order to achieve robust and accurate matching, epipolar lines are calculated and used for refinement of matching. The experimental results show that the simplified algorithm can not only reduce the time complexity, maintain robust quality, extract enough feature points to match, but also improve the image matching efficiency. [Copyright &y& Elsevier]

Abstract: Based on the theory of wavelet video image compression and according to the feature of wavelet transformation and applying to the key technology of wavelet video processing, this paper has analyzed the existing wavelet video image processing technology and tried to point out the new location of multi-level transformation arrangement algorithm. The algorithm has great advantages over the existing compression algorithm. The experimental results show that this convolution algorithm has a better compression ratio. [Copyright &y& Elsevier]

Abstract: The prediction of intonational phrase (IP) breaks is important for both the naturalness and intelligibility of Text-to- Speech (TTS) systems. In this paper, we propose a maximum entropy (ME) model to predict IP breaks from unrestricted text, and evaluate various keyword selection approaches in different domains. Furthermore, we design a hierarchical clustering algorithm for automatic generation of feature templates, which minimizes the need for human supervision during ME model training. Results of comparative experiments show that, for the task of IP break prediction, ME model obviously outperforms classification and regression tree (CART), log-likelihood ratio is the best scoring measure of keyword selection, compared with manual templates, templates automatically generated by our approach greatly improves the F-score of ME based IP break prediction, and significantly reduces the size of ME model. [Copyright &y& Elsevier]

Abstract: Traditional algorithms in hiding sensitive association rules do not consider the correlation between rules, which lead to a high ratio of hiding failure and a big side effect. We propose a new privacy preserving algorithm for sensitive association rules based on least transactions to be modified in this paper, which considers the correlation between sensitive rules sufficiently. The proposed algorithm first creates a table in the main memory, then for each sensitive rule, inserts all the information about modification rules which can decrease their support or confidence into the table. After that, it compares all the modification rules and removes the modification rules which will increase the support or confidence of some sensitive rules. Then the algorithm selects transactions which satisfy maximum modification rules characteristics to modify each time. Thus, it will need much less transactions to be modified overall. Theoretical analysis and the experiment results show that the proposed algorithm can not only reduce the ratio of hiding failure but also modify much less transactions, and it produces a small side effect. [Copyright &y& Elsevier]

Abstract: Traditional fuzzy clustering algorithms have poor effects on nonlinearly separable data sets. In this paper, a kernelbased possibilistic fuzzy c-means clustering algorithm is proposed. Through Mercer kernel functions, the samples in the original space are firstly mapped into a high-dimensional feature space. Then the samples in the feature space are clustered using the possibilistic fuzzy c-means clustering algorithm. This method is not only effective on the linearly separable data sets and nonlinearly separable data sets, but also robust to noises. The experimental results show the effectiveness and feasibility of this algorithm. [Copyright &y& Elsevier]

Abstract: This paper presents a new keyword extraction algorithm for Chinese news web pages which is bases on the traditional TFIDF method and improves it by combining channel division within and putting word co-occurrence feature into consideration. Word co-occurrence distribution is an important statistical model widely used in natural language processing that reflects the co-relationship of the words. And the channel division can improve the accuracy of IDF value so as to improve the efficiency of TFIDF algorithm. Experiments on randomly selected web pages have been performed to demonstrate the quality of the keywords extracted by our proposed algorithm. [Copyright &y& Elsevier]

Abstract: In this paper, the Multi-Depot Vehicle Routing Problem (MDVRP), a generalization of the Vehicle Routing Problem (VRP) is reviewed. We mainly review the MDVRP from the following several aspects: the definition, solution methods (including exact algorithms, heuristics and meta-heuristics) and problem variants. Our objective is to survey the MDVRP systematically and draw more attention to it from researchers and practitioners. [Copyright &y& Elsevier]

Abstract: A new classification algorithm of M-ary FSK signals based on the extraction of signals’ features is presented in this paper. The implementation of the methodology does not require any apriori information. By extracting some basic features of intermediate frequency signals both in time domain and frequency domain, the classification of M-ary FSK signals is achieved. Simulation result shows that the proposed algorithm can classify 2FSK, 4FSK and 8FSK signals effectively under AWGN and and still keeps a good performance even at low input SNR. [Copyright &y& Elsevier]