Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-04-01 Cooperative journals: 《计算机应用研究》
Abstract： For the problem of the palmprint recognition algorithm had less research on the recognition of color palmprint images, this paper proposed a new BP neural network color palmprint image recognition algorithm based on the analytical properties of Stein-Weiss function. The proposed method first constructs a Stein-Weiss function for each pixel in the color palmprint image. Then according to the analyticity of the Stein-Weiss function, sixteen eigenvalues of the corresponding pixel are calculated. These eigenvalues are used as the input data of the BP neural network. These data can be classified and learned through the self-learning ability of BP neural network. Then the palmprint edge is obtained through generalization ability of BP neural network. Finally, the pairwise attributes are extracted from palmprint edge, and the feature library is built. The palmprint is recognized by the paired geometric histogram intersection algorithm. The experimental results show that the proposed algorithm can extract finer palmprint faster and the recognition rate is higher compared with the previous grayscale palmprint image recognition algorithm. And it has strong robustness to the interference of rotation and noise.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-04-01 Cooperative journals: 《计算机应用研究》
Abstract： Considering the unique users of scholars' social networks, this paper proposes a collaborative association model of paper and project data based on scholars' social networks. Firstly, The proposed model uses the two-step feature selection method to preprocess the data, while removing the irrelevant and redundant features. So that the model would obtain the effective features that affect the association between the paper and the project. Then it would adopt text vector space model to calculate the text similarity between the paper and the project. After finishing these, it could form recommendation sets for different papers/projects. Through the social network "SCHOLAT" data for researchers, the model is implemented and applied to SCHOLAT. The online application situation and user feedback show that the model has good accuracy and practicability. Furthermore, it can more fully explore the potential relationship between the paper and the project, provide users with better academic research management services, and propose a novel research method for analyzing the academic big data.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-04-01 Cooperative journals: 《计算机应用研究》
Abstract： The problem of minimizing the maximum interference in wireless sensor networks can be described as follows: given n points with their positions on the plane and thresholds of transmission ranges , for each point, the maximum number of other points whose transmission ranges cover this point is minimized. To effectively solve this problem, this paper proposed a new greedy algorithm called edge shrinking algorithm. The algorithm constructs a desirable network communication topology by shrinking edges, which is different from existing algorithms and is accelerated by incorporating the idea of batch processing in operating system to shorten its running time. The result shows that this algorithm can produce better solutions than existing algorithms.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： TTL is a highly efficient indexing structure for finding an earliest arrival path, or a latest departure path, or a shortest duration path in public transportation networks. TTL uses Time-dependent Dijkstra’s algorithm as its core algorithm to build index, and is therefore, results in two deficiencies. Firstly, it needs relatively expensive priority queue operations. Secondly, it would generate paths with more transfers. This paper proposes a new indexing structure, TAIL, which uses a trip based method to build index. TTL pre-computes some canonical paths. A query could be answered by matching up the canonical paths, which avoids traversing the entire network. Instead of the graph structure, TAIL uses trip array as its input, and generates paths by scanning trips. Initially, TAIL scans trips starting from the source stop, from which TAIL obtains direct reachable stops. After that, TAIL scans trips starting from the direct reachable stops, from which TAIL obtains reachable stops within one transfer. Generally, TAIL discovers new reachable stops from scanning the trips starting at the already discovered reachable stops. In order to obtain the earliest arrival paths in the early stage, so as to reduce the number of trip scanning, TAIL does not scan the stops strictly in increasing order of their transfer times. In this way, TAIL avoids valuable priority queue operations, while preserves the entity of a trip. Experiments on real datasets shows that, compared to TTL, TAIL has lower index construction time and its generated path has fewer transfer times.
Hits
Hits
Downloads
Comment