Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-08-13 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the problem that there exists item deviation and user preferences in collaborative filtering recommendation for the interaction between factors related in items and users, this paper proposed a recommendation algorithm integrated item deviation and user preference. Firstly it clustered to generate item clusters on LDA topics modeling and to get user clusters by using K-means; then it generated item deviation score on the constraints of item cluster and user cluster, and obtained user preference score with probability transfer on user-item score and item type. Finally it weighted the item deviation score and user preference score linearly to form the prediction score based on the existing scoring average in the item cluster. Comparison experiments show that the new algorithm could obtain reasonable recommendation based on different neighbors and improve recommendation accuracy.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-06-19 Cooperative journals: 《计算机应用研究》
Abstract： The texture revealed by line integral convolution (LIC) reflects the directional structure of the whole vector field, but cannot show the field intensity. In order to solve this problem, this paper proposed an improved LIC algorithm based on nonlinear gradual-changing color mapping. Firstly, the intensity of vector field combined with white noise to form the input texture of improved LIC, which used the idea of FastLIC and divides the texture into several regions for synchronous execution of LIC, to improve the efficiency of LIC. Secondly, the improved LIC does a non-linear transformation on the intensity of vector field, and used the processing engine, OpenCV, to implement the color mapping of the vector field intensity according to the gradual-changing color-mapping scheme. Finally, the results of gray texture obtained by LIC and color mapping determine the synthetic coefficient, constructing a cumulative function enhances the two results, and adopting the way of linear synthesis obtains the final visualization. Simulation results show that when applied to the global ocean flow field and wind field, the proposed algorithm generates a clear visualization displays the direction and intensity of vector field better, and reflects the global information and local changes of the vector field better, when compared with other algorithms.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-06-19 Cooperative journals: 《计算机应用研究》
Abstract： B-spline curve fitting is applied to draw the changing trend of discrete data points, which usually obtains by data approximation or iterative method. It plays an important part in image processing and reverse engineering. Aiming at the situations where multi peak, cuspidal point or discontinuity exists in the curve to fit, this paper proposed a B-spline curve fitting algorithm based on genetic algorithm. Firstly it used the penalty function to transform the constrained optimization problem into an unconstrained problem. Then it used an improved genetic algorithm to select an adaptive fitness function, and adjusted the number and positions of nodes adaptively by combining the simulated annealing algorithm to find the optimal node vector. The iterations continued until generating the final good reconstruction curve. Experimental results show that the algorithm improves accuracy and speeds up convergence effectively.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-24 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the negative effects of makeup on face recognition, this paper presents an improved robust face recognition algorithm based on patch integration learning. In the pro-posed scheme, this paper tessellates each face image into patches and represents each patch with a set of feature descriptors, viz. , Local Gabor Pattern (LGP) , Histogram of GaborSpaceFixed Ratio Measures (HGSFRM) and Densely Sampled Local Multi-valued Pattern (DSLMP) . Then, uses an improved Random Subspace Linear Discriminant Analysis (SRS-LDA) method to perform ensemble learning by sampling patches and constructing multiple common subspaces between before-makeup and after-makeup facial images. Finally, uses Collaborative- based and Sparse-based Representation Classifiers to compare feature vectors in this subspace and the resulting scores combined by the sum-rule. The proposed face matching algorithm is evaluated on multiple makeup dataset. Results show that this algorithm has higher recognition accuracy than other algorithms designed for face recognition under makeup situation.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： The high sampling rate of GPS makes the data set of the trajectory huge, which is difficult to handle in practical applications. It needs to rely on trajectory simplification algorithm to compress the original data. Aiming at this problem, a new simplification algorithm based on speed segmentation was proposed, namely STS algorithm, which preserved the spatiotemporal characteristics while preserving the velocity characteristics of a given trajectory. The STS algorithm divided the velocity values into several intervals and divides the trajectory into velocity-preserving segments. It calculated the SED threshold of each trajectory segment and derives a simplified trajectory by applying the TD-TR algorithm on each sub-trajectory segment. Extensive experiments with real datasets demonstrate that the proposed algorithm has better performance than ATS algorithms.
Hits
Hits
Downloads
Comment