In the indoor robot localization and 3D dense mapping, the existing methods can not satisfy the requirements of high-precision localization, large-scale and rapid mapping. The ORB-SLAM (Oriented FAST and Rotated BRIEF-Simultaneous Localization And Mapping) algorithm, which has three parallel threads including tracking, map building and relocation, was used to estimate the three-dimensional (3D) pose of the robot. And then 3D dense point cloud was obtained by using the depth camera KINECT. The key frame extraction method in spatial domain was introduced to eliminate redundant frames, and the sub-map method was proposed to reduce the cost of mapping, thereby the whole speed of the algorithm was improved. The experiment results show that the proposed method can locate the robot position accurately in a large range. In the range of 50 meters, the root-mean-square error of the robot is 1.04 m, namely the error is 2%, the overall speed is 11 frame/s, and the localization speed is up to 17 frame/s. The proposed method can meet the requirements of indoor robot localization and 3D dense mapping with high precision, large-scale and rapidity.

According to the problem of excessive energy consumption caused by the unreasonable distribution of cluster head nodes in the large-scale Wireless Sensor Network (WSN), a Hierarchical Routing Protocol for wireless sensor networks based on Non-uniform Clustering (HRPNC) was designed. HRPNC combined the idea of clustering in Low Energy Adaptive Clustering Hierarchy (LEACH), and basing on stratification improved the algorithms of competitive radius regarding Energy-Balanced Unequal Clustering routing protocol for WSN (DEBUC). Through taking advantage of hierarchical mechanism and the mechanism of competition, the distribution of the cluster heads turned out to be more reasonable and the energy consumption of such nodes got balance effectively. In the simulation performed on the Matlab, the life cycle of HRPNC was higher than that of the LEACH and DEBUC by about 500 and 300 rounds respectively. The average residual energy of the nodes with HRPNC was higher than that of the nodes with LEACH and DEBUC. As to the energy consumption, it remained lower and more stable during the survival phase. Besides, compared with LEACH and DEBUC, the aggregate of data packet of HRPNC was 300% and 130% higher respectively. What is more, under different simulations, the packet loss rate of HRPNC was lower than that of LEACH and DEBUC. The experimental results show that HRPNC can not only extend the lifetime of the network, and increase network stability and the number of data transmission, but also reduce the loss rate of data transmission effectively.

Concerning the problems of the low accuracy of community detection in bipartite network and the strong dependence on additional parameters, depending on the original network topology, based on the idea of spectral clustering algorithm, a new community algorithm was proposed. The proposed algorithm mined community by mapping a bipartite network to a single network, substituted resource distribution matrix for traditional similarity matrix, effectively guaranteed the information of the original network, improved the input of spectral clustering algorithm and the accuracy of community detection. The modularity function was applied to clustering analysis, and the modularity was used to measure the quality of community mining, effectively solved the problem of automatically determining the clustering number. The experimental results on the actual network and artificial network show that, compared with ant colony optimization algorithm, edge clustering coefficient algorithm etc., the proposed algorithm can not only accurately identify the number of the communities of the bipartite network, but also obtain higher quality of community partitioning without previously known parameters. The proposed algorithm can be applied to the deep understanding of bipartite network, such as recommendation and influence analysis.

In E-commerce website, massive disorder shopping reviews may make the consumers be lost in the massive shopping reviews and can not distinguish trusted reviews. Therefore, this paper proposed a trustworthy sort method for customer reviews. Firstly, focusing on commercial advertising information in websites and concerning about whether the contents of the online customer reviews and product functional properties are closely related, the authors designed an algorithm of product's key features extractions from shopping websites based on HTML script format, and presented a method of customer reviews features extractions based on natural language processing. Secondly, the authors used the technique of words similarity to analyze the correlation degree between product features and customer reviews contents, and then proposed the computational method of trust degree for shopping customer reviews. Finally, through analyzing the method with an example, the proposed method achieves a trustworthy sort for large online shopping customer reviews. Thus customers need not browse all reviews to judge which one can be trusted or have the real reference value. It decreases information search costs and improves the efficiency of decision making.

According to the fact that the performance of skin-color detection is greatly affected by the illumination, a kind of skin-color detection algorithm with good stability was proposed. According to the characteristic of face symmetry, the pixel correction algorithm was used to replace too bright or too dark pixels on the face area with normal ones, and then an adaptive method was used for skin-color detection, in which the corresponding chroma threshold was determined dynamically by the brightness of pixel. The experimental results show that, compared to other algorithms such as the YCbCr single Gauss model for skin-color detection, more than 10% of positive detection rate was increased and the false positive rate was reduced by 5% with the proposed algorithm under different light intensity. Moreover, the stability of the proposed algorithm is significantly enhanced.

This paper proposed a method for skin segmentation based on the similarity of skin-tone given by strengthened classifier which constructed via Real AdaBoost algorithm and dynamic threshold.Based on the clustering property of skin-tone distribution in YCrCb chrominance space, a set of weak classifiers in Look-Up-Table (LUT) type using circle-like features was trained via Real AdaBoost to form a strengthened classifier. Firstly, gray scale images indicated the skin-tone similarity of the pixels was created by processing the images with the strengthened classifier. Then skin segmentation was implemented according to the dynamic threshold selected through Da-Jing method. The experimental results show that the strengthened classifier has an outstanding ability for describing the distribution of skin-tone color in the YCrCb space. The method is robust, and efficient.