Fast and reliable hand pose estimation has a wide application in the fields such as human-computer interaction. In order to deal with the influences to the hand pose estimation caused by the light intensity changes, self-occlusions and large pose variations, a deep network framework based on label distribution learning was proposed. In the network, the point cloud of the hand was used as the input data, which was normalized through the farthest point sampling and Oriented Bounding Box (OBB). Then, the PointNet++ was utilized to extract features from the hand point cloud data. To deal with the highly non-linear relationship between the point cloud and the hand joint points, the positions of the hand joint points were predicted by the label distribution learning network. Compared with the traditional depth map based approaches, the proposed method was able to effectively extract discriminative hand geometric features with low computation cost and high accuracy. A set of tests were conducted on the public MSRA dataset to verify the effectiveness of the proposed hand pose estimation network. Experimental results showed that the average error of the hand joints estimated by this network was 8.43 mm, the average processing time of a frame was 12.8 ms, and the error of pose estimation was reduced by 11.82% and 0.83% respectively compared with the 3D CNN and Hand PointNet.

Shape registration is a common task in non-rigid 3D shape analysis. In order to solve the problems of high complexity, large computation cost and low accuracy of the traditional algorithms, a new hierarchical shape registration method was proposed. Firstly, the heat kernel signature function was defined as the scalar field for a model, and persistence-based clustering was used to extract feature points and salient regions of the model. Then, a novel tree-based shape representation was proposed, whose root node, internal nodes and leaf nodes were defined as the model, the salient regions and the corresponding feature points, respectively. Finally, a new hierarchical shape registration method was designed to make full use of the tree-based shape representation. The hierarchical shape registration algorithm was tested on the SHREC 2010 correspondence dataset and compared with the Generalized Multi-Dimensional Scaling (GMDS) and game theory algorithms. Experimental results show that the proposed hierarchical shape registration method achieves higher accuracy than GMDS and game theory under various shape transformations, including isometric transformation, holes, micro holes, scaling, local scaling, resampling, noise, shot noise and topological transformation; in addition, the computational complexity is reduced significantly.

A large amount of alarms considerably complicate the root-cause analysis in telecom networks, thus a new alarm filtering algorithm was proposed to minimize the interference on the analysis. Firstly, a quantitative analysis for the alarm data, e.g., the quantity distribution and the average duration, was conducted, and the concepts of alarm impact and high-frequency transient alarm were defined. Subsequently, the importance of each alarm instance was evaluated from four perspectives:the amount of the alarms, the average duration of the alarms, the alarm impact, and the average duration of the alarm instance. Accordingly, an alarm filtering algorithm with O ( n) computation complexity in principle was proposed, where n is the number of alarms under analysis. Single-factor experimental analysis show that the compression ratio of the alarm data has a positive correlation with the alarm amount of a specific alarm element, the average duration of the alarms, the alarm impact, and the duration of the alarm instance; further, the accuracy of the proposed algorithm is improved by 18 percentage points at most compared with Flexible Transient Flapping Determination (FTD) algorithm. The proposed algorithm can be used both for off-line analysis of historical alarm data and for on-line alarm filtering.

Regarding to the resource allocation and partition starting time in the Integrated Modular Avionics (IMA), a Unified Modeling Language (UML) model of partition configuration and initialization was proposed based on the case study of VxWorks 653 partition operating system. The proposed model including classes diagram and initial sequence diagram for partition, was established to facilitate the analysis of the mechanism of partition configuration and starting/initialization. The contents and function of partition configuration in the processes of resources allocation, operating system compilation and partition initialization, were discussed in detail, as well as the differences between "cold start" and "warm start" mode. A platform was set up for testing the startup times of the two kinds of startup modes, and the test results showed that the time of cold start was 148 ms, and warm start time was 8.5 ms. Furthermore, the applicable scenarios for cold start and warm start mode were discussed. The policies of partition configuration and application software initialization were proposed based on the starting time. The mode of partition start and time of partition initialization should be fully considered when establishing the partition main time frame and identifying the health management policy. The designed policies can be applicable to other partition system design in high security applications.

In recent years, the complexity of conducting schedulability experiments increases with the rapid development of real-time scheduling research. In general, schedulability experiments are time-consuming in the absence of standardized and modularized experiment tools. Moreover, since the source codes are not publicly available, it is difficult to verify the reported results in the literature, and to reuse and extend the experiments. In order to reduce the repeative work and help the vertification, a basic schedulability experiment framework was proposed. This experiment framework generated task systems through random distribution, and then tested their schedulability, and based on the framework, a novel open-source schedulability platform called SET-MRTS (Schedulability Experiment Toolkit for Multiprocessor Real-Time Systems) was designed and realized. The platform adopted the modular architecture. SET-MRTS consisted of the task module, the processor module, the shared resource module, the algorithm library, the configuration module and the output module. The experimental results show that, SET-MRTS supports uni- and multi-processor real-time scheduling algorithms and synchronization protocol analyses, which can correctly perform the schedulability test and output intuitive experimental results, and support the expansion of the algorithm library. Compared with algorithms in the algorithms library implemented in the experiment, SET-MRTS has good compatibility and expansibility. SET-MRTS is the first open source platform to support a complete experimental process, including algorithmic implementation, parameter configuration, result statistics, charting, and so on.

In recent years, Convolutional Neural Network (CNN) has made a series of breakthrough research results in the fields of image classification, object detection, semantic segmentation and so on. The powerful ability of CNN for feature learning and classification attracts wide attention, it is of great value to review the works in this research field. A brief history and basic framework of CNN were introduced. Recent researches on CNN were thoroughly summarized and analyzed in four aspects: over-fitting problem, network structure, transfer learning and theoretic analysis. State-of-the-art CNN based methods for various applications were concluded and discussed. At last, some shortcomings of the current research on CNN were pointed out and some new insights for the future research of CNN were presented.

Widely used WebKit of excellent architecture has been ported to many embedded platforms, with excellent cross-platform features. Due to the diversity of hardware for embedded platforms, WebKit open source version does not take full advantage of the characteristics of embedded platforms. Through studying WebKit render architecture, taking full advantage of the embedded hardware feature and the benefit of the software rendering design, a hardware and software fusion render layer was designed. This layer sped up the browser rendering on the embedded platform and improved the user experience. The layer was verified, the time of opening website was reduced by 48% and the speed of rendering of html animation increased by 130% compared to the original WebKit.

The common computing resources in the Integrated Modular Avionics (IMA) system provide the hosted applications with temporal and spatial partitioning platform. The platform for applications development should comply with the ARINC 653 specification which defines the application executive interfaces for partitioning operation system. By porting and developing the Board Support Package (BSP) and AFDX network driver, for the first time, an IMA platform solution for hosted application development was achieved based on the C2K, a Commercial Off-The-Shelf (COTS) single board computer. The functionality and performance of the COTS based platform are similar to the popular common computing resources in IMA of modern civil transportation aircraft, offering a development platform for hosted applications development and debug at a pretty lower costs.

A parallel parsing and layout method for embedded browser was introduced in this paper. By putting the controlling and layout into the master thread, loading resources and parsing into the slave thread, this method paralleled the serial parsing and layout process for Web pages. To solve the problem of the browser’s slow response to users, the message scheduling method for the browser combined with thread scheduling model in Operating System (OS) was also designed. The feasibility of the method is verified and the speed of rendering the first screen to users is improved by 4%-17% costing less than 64MB memory according to the testing cases.

Due to the variable interdependencies among the operations and preconditions, certain paths of an extended finite state machine (EFSM) may not be feasible. These infeasible paths complicate the class test generation process from EFSMs. By transforming UML state diagrams into extended finite state machines (EFSMs), an algorithm is presented to eliminate the infeasible paths in these EFSMs. Thus a class test model for object-oriented software can be constructed.The model is suitable for class testing by using control and data flow analysis techniques.