Search

Effectively learning and extracting the feature representations of 3D point clouds is an important yet challenging task. Most of existing works achieve reasonable performance in 3D vision tasks by modeling the relationships among points appropriately. However, the feature representations are only learned with a specific transform through these methods, which are easy to overlap and thus limit the representation ability of the model. To address these issues, we propose a novel Multi-Transform Learning framework for point clouds (PointMTL), which can extract diverse features from multiple mapping transform to obtain richer representations. Specifically, we build a module named Multi-Transform Encoder (MTE), which encodes and aggregates local features from multiple non-linear transforms. To further explore global context representations, a module named Global Spatial Fusion (GSF) is proposed to capture global information and selectively fuse with local representations. Moreover, to guarantee the richness and diversity of learned representations, we further propose a Spatial Independence Criterion (SIC) strategy to enlarge the differences between the transforms and reduce information redundancies. In contrast to previous works, our approach fully exploits representations from multiple transforms, thus having strong expressiveness and good robustness for point clouds related tasks. The experiments on three typical tasks (i.e., semantic segmentation on S3DIS and ScanNet, part segmentation on ShapeNet and shape classification on ModelNet40) demonstrates the effectiveness of our method.

The intermittence and uncertainty of wind power pose challenges to large-scale wind power grid integration. The study of wind power uncertainty is becoming increasingly important for power system planning and operation. This paper proposes a wind power probabilistic interval prediction model, and a novel reliability assessment approach is presented for electrical power systems. First, the unknown parameters estimation of the autoregressive integrated moving average (ARIMA) prediction model is based on the Markov chain Monte Carlo (MCMC)-based Bayesian estimation method to improve the quality of statistical inference. Then, a quantum genetic algorithm is used to segment the power to determine the best output for each power segment weight and calculate the probabilistic prediction interval of wind power. Finally, reliability assessment by the sequential Monte Carlo simulation is presented combining with the probabilistic prediction interval of wind power on IEEE-RTS79 reliability test system. The simulation results that proposed variation range of reliability assessment indices consider the uncertain scenario of wind power and has guiding significance for power generation scheduling. Compared with genetic algorithm and particle swarm optimization algorithm, it is proved that the proposed prediction interval model has better prediction interval coverage probability index and interval average bandwidth index.

In order to solve the problem of hopping mechanism design of robot, we proposed a method of optimization of transmission mechanism based on parametric model. In this paper, the optimization of energy transfer efficiency of robot jump mechanism is completed by parametric model method through the mechanism simplification, parametric modeling and parameter optimization. The results show that the method can greatly optimize the energy transfer efficiency and reduce the driving torque required by 74.6%. It provides a feasible optimization method for the design of the robot hopping structure.

Aiming at the problem of poor trafficability and difficulty in dealing with complex environment of traditional mobile robot, we proposed a climbing and hopping hybrid mobile robot. In this paper, we took the spring four-bar linkage mechanism as the hopping mechanism and the C-shaped wheel foot mechanism as the walking mechanism. Completed the structure design of the robot. Realized the assembly and disassembly of the virtual prototype, and completed the motion simulation and analysis. The results show that the robot is stable in the aspects of structure design and its jumping height can reach 988mm, and has a great potential in reconnaissance, monitoring, survey etc.

This paper proposes a new method to generate pseudo-annotations from manual bounding boxes for semantic segmentation. Different from traditional local data driven based methods such as Conditional Random Field (CRF) and GrabCut, we aim at using class-agnostic bounding box based segmentation models. To this end, we propose a new segmentation network, which formulates segmentation task as a sparse boundary point detection task rather than dense pixel label prediction task, and therefore can provide new type of pseudo-annotations. Furthermore, we detect object boundary based on direction, and use multiple directions to handle various shapes of objects. Moreover, we further enhance the pseudo generation by combining different types of segmentation masks. Classical Fully Convolutional Networks (FCN) network based on dense prediction is also modified to generate diverse foreground masks. A simple fusion method based on intersection operation is proposed to combine the two types of pseudo-annotations. We verify the effectiveness of our method on PASCAL VOC 2012 validation dataset. The mIoU value is 67.9%, which outperforms the state-of-the-art method by 1.1%.

This paper proposes a digital watermarking algorithm for color image based on Dual-Tree Complex Wavelet Transform (DTCWT). It decomposes the color watermark image M in the RGB color space into three channels, i.e. R, G and B, and then applies Arnold transform on them respectively. It then applies DCT (Discrete Cosine Transform) and DFT (Discrete Fourier Transform) on the scrambled watermarks in the frequency domain. The algorithm applies Dual-Tree Complex Wavelet Transform (DTCWT) on the color carrier image, and uses the third-level low-frequency sub-bands of its real part as the embedded sub-bands. It then embeds all the above-mentioned watermarks into these sub-bands. Finally, it applies inverse Dual-Tree Complex Wavelet Transform on the color carrier image embedded with the color watermark to embed the color watermark into it. Watermark extraction is the inverse process of watermark embedding. Experimental results show that this algorithm features high imperceptibility and robustness.

3D reconstruction using multiple views allows for the restoration of a complete geometric model, but does not produce textural effects. Most of the existing texture mapping methods are aimed at the scanning reconstruction models, and few of them can be used to deal with the models with large amount of data and complex structure. We propose a high-realistic texture mapping algorithm based on image sequences. Firstly, the image sequence is sampled by using the spatio-temporal adaptive method. Then, the parameters of the camera and the size of each image are extracted by means of the Bundler, and the optimal texture image is selected for each triangular patch through the Markov random field. Finally, due to excessive loading of image data, it is necessary to reduce the texture data and only retain the effective parts mapped onto each triangular patch. Our method ensures that the resolution of the model after texture mapping is higher, and there is no obvious texture seam.

A 3D point cloud model reconstructed based on image usually contains many outliers. These outliers may exist in the form of isolation or aggregated clusters, or connected to surface of model. Based on the distribution characteristics of outliers, an integrated outlier detection method is proposed. Firstly, the spatial topological relation of point cloud data is established by dividing cell method. When the surface of model sampling is insufficient, the points of surface will be split into a number of clusters with different sizes instead of a complete cluster, the boundary matching method can be used to preserve the good points and remove outlier clusters. Finally, the improved K-means algorithm is adopted to cluster all points based on colors, and remove the outliers which are connected to surface. Simulation results show that the proposed algorithm can effectively detect outliers with the different distribution state in the reconstructed point clouds.

Accidents caused by defective software systems have long been a nightmare. Though engineers utilize advanced techniques and rigorous quality control procedures, we still have to admit that the increasing complexity and expanding scale of software systems make it extremely difficult to guarantee high quality deliverables. Since large-scale software systems exhibit the characteristics of complex networks, applying the principles of complex networks to evaluate the quality of software systems has attracted attention from both academia and industry. Unfortunately, most current research studies focus only on one or a limited number of attributes of software structures which makes them ineffective in providing comprehensive and insightful quality evaluation for software structures. To overcome this problem, we propose an approach based on various software structural characteristics to evaluate software structures from modularity, hierarchy, complexity, and fault propagation points of view. A model based on these four aspects is proposed to better understand software structural quality. A prediction model is also proposed to provide insights on the nature of software evolution and its current status. Experiments using two software projects were performed against the thresholds obtained by evaluating more than 5,000 versions of open source projects. Our results suggest that the approach described in this paper can help us analyze real-world software projects for better quality evaluation.

Publish/Subscribe (Pub/Sub) systems are the middleware which provides the distributed event detection. The increasing requirements from applications have driven Pub/Sub systems to provide diversified quality of service (QoS). Therefore, this paper builds a Pub/Sub system Phoenix on the cloud which can meet the different needs of system reliability and user timeliness. In brief, Phoenix can deal with link failure, broker failure and cluster churning to guarantee that the system can run normally. Meanwhile, it provides the timeliness guarantee mechanism to manage to satisfy the timeliness specified by the users with its best effort. The experimental results show that Phoenix is fairly reliable and accommodates the users' personalized timeliness requirements well.

A spatial 2-RPU&SPR parallel manipulator (PM) with three degrees of freedom (DOF) is proposed in this paper. The architecture of the manipulator is comprised of a moving platform attached to a base through two revolute-prismatic-universal jointed serial linkages and one spherical-prismatic-revolute jointed serial linkage. The prismatic motions of the prismatic joints are considered to be actively actuated. The moving platform of the manipulator has one-translational and two rotational (1T2R) DOFs with respect to the fixed base. The kinematics including forward and inverse position analysis is analyzed. At last, the simulations of position and velocity of the manipulator are made to validate the algorithm.

With the expanding wind power scale by each year, the traditional method for optimizing spinning reserve (SR) capacity is no longer applicable. Due attention has to be paid on this added wind uncertainty when setting the SR requirement. For obtaining the more optimal and reasonable results, a technique that SR determination can be dealt should be researched in wind power integration system. Therefore, with involving wind farm output variables, the fresh and upgrade Robust Linear Optimization (RLO) which proposed by Seong-Cheol Kang is firstly introduced into constructing the model of minimum generating the cost of conventional generators (CGs) and SR cost and its optimal capacity in this paper. The wind farm output was described as its minimum, maximum and mean values in the model through the SCK robust linear optimization theory. The reliability for optimal SR capacity and total cost can be preliminarily calculated under a variety of wind power fluctuation in this model. The validity of the proposed method will be verified by the test results of the IEEE-30 bus system. The relationships of wind uncertainties and required SRs can also be confirmed.

Content-based Pub/Sub systems enable various applications especially in the field of Internet of Things (IoT) to capture the events interested by the users. However, the event/subscription routing approaches in the Pub/Sub systems must adapt to the diversity and skew ness of the events/subscriptions. Only on this basis, performance improvement and quality of service required by the users can be considered. In the face of changing event/subscription loads, this paper presents an adaptive approach to channel management in the channel-based routing. Our approach exploits the historical information of the subscriptions and events as well as the covering relationship between subscriptions to accomplish channel splitting and joining. Due to properly utilizing the differencing method of set partitioning, our approach can guarantee that the difference between loads of two resultant brokers is as small as possible at the granularity of a subscription tree after a channel splitting operation on an overloaded broker. Moreover, it can work with any content-based subscription language without any limitation imposed on the specifications of subscriptions. The experimental results show that our approach is adaptable and elastic in terms of standard deviation of loads of brokers and also has acceptable offloading time.

The character of reactive power in wind power integrated system affects active power losses and voltage level of the entire system. For the influence from the randomness of wind speed to entire system power flow, a reactive power optimization model based on Robust Optimization (RO) is presented firstly. With minimizing power losses as an objective, the uncertainty of the wind speed is taken as a random variable and demonstrated in a "box" uncertainty set for transforming the uncertain constraints into the certain form. The duality theorem is utilized to simplify the model to be general linear programming ones. IEEE30-node system is used as an example to test the model and method. The results show that the random wind speed is quantitated effectively, power losses are reduced, and even the voltage of entire system can be restricted within regular range. The reactive power optimization method with RO is simple, feasible and significant.

Combined Cold Heat and Power (CCHP) is of great advantage in energy cascade using, but the energy grade brings more difficulties for energy equal transformation. Considering the rising of energy transformation efficiency, and the coefficient is decreasing which power converts into coal, so the transformation of different energy will change with production status. So we set out to research the problem from the angle of pollutant emission based on heat equivalent (HE), and proposed the principle of equal emission (PEE) that can reflect the dynamic transformation between cold heat and power, and establish the dynamic transformation model of different energy. Thereafter, the model of multi-objective economic dispatch is formulated including generation cost and environmental cost of CCHP based on the PEE, and then we solve the problem with the fuzzy optimization programming. The results show that PEE is important in economic dispatch and reducing emission, the more important is that it can reflect the change of energy production more accurately; it also provides the reference for the promotion of environmental dispatch of CCHP.

Pub/Sub services in the cloud are intended to achieve high performance and high scalability. The challenge comes from how to deal with the potential wide variation in workloads. The paper constructs a content-based Pub/Sub service named OPS4Cloud on the top of cloud infrastructure from scratch. OPS4Cloud manages its brokers by regions and presents a VM-based region adjustment strategy. It builds two-level indexes at brokers to speed up the routing of subscriptions, events and advertisements on the basis of channelization routing. It also designs a mechanism for subscription persistence, accelerating the load transfer. The paper conducts extensive experiments on OPS4Cloud, especially in comparison with the migrated Pub/Sub systems. The experimental data shows that OPS4Cloud is vastly superior to the migrated systems in terms of system throughput and event response time, and it also tends towards a load-balancing state among brokers while running continuously.