Loading...

Table of Content

    30 October 2018, Volume 24 Issue 5
    Digital Film and Television Technology
    Hand segmentation from a single depth image based on histogram threshold selection and shallow CNN
    XU Zhengze, ZHANG Wenjun
    2018, 24(5):  675-685.  doi:10.12066/j.issn.1007-2861.2073
    Asbtract ( 902 )   HTML ( 178)   PDF (2977KB) ( 256 )  
    Figures and Tables | References | Related Articles | Metrics

    Real-time hand gesture recognition technology significantly improves the user's experience for virtual reality/augmented reality (VR/AR) applications, which relies on the identification of the orientation of the hand in captured images or videos. A new three-stage pipeline approach for fast and accurate hand segmentation for the hand from a single depth image is proposed. Firstly, a depth frame is segmented into several regions by histogram-based threshold selection algorithm and by tracing the exterior boundaries of objects after thresholding. Secondly, each segmentation proposal is evaluated by a three-layers shallow convolutional neural network (CNN) to determine whether or not the boundary is associated with the hand. Finally, all hand components are merged as the hand segmentation result. Compared with algorithms based on random decision forest (RDF), the experimental results demonstrate that the approach achieves better performance with high-accuracy (88.34% mean intersection over union, mIoU) and a shorter processing time ($\le $8 ms).

    Scratch repair algorithm based on 5${\times}$5 neighborhood pixels correlation
    XU Min, DING Youdong, YU Bing, LI Chang, WU Biao, ZHANG Wanying
    2018, 24(5):  686-693.  doi:10.12066/j.issn.1007-2861.2083
    Asbtract ( 1097 )   HTML ( 16)   PDF (13172KB) ( 289 )  
    Figures and Tables | References | Related Articles | Metrics

    In view of the existence of gradient effect and low accuracy of traditional scratch repair algorithm, this paper proposes a new scratch repair algorithm based on 5$\times$5 neighborhood pixels correlation, using the correlation between the repaired area and the surrounding 24 adjacent pixels, and determining the filling order by defining the priority of the repaired pixels. Each time a pixel is repaired and the area to be repaired is updated in time as well. This operation is repeated until the number of the pixels in the region to be repaired comes to 0. The experimental results show that compared with other traditional algorithms, the algorithm is more suitable for repairing old film and television materials, which not only shortens the running time, but also obtains higher peak signal to noise ratio (PSNR) and structural similarity index (SSIM) values.

    Global path planning of film group animation based on improved JPS algorithm
    HUANG Dongjin, LEI Xue, JIANG Chenfeng, CHEN Yanmin, DING Youdong
    2018, 24(5):  694-702.  doi:10.12066/j.issn.1007-2861.2074
    Asbtract ( 1190 )   HTML ( 17)   PDF (13202KB) ( 320 )  
    Figures and Tables | References | Related Articles | Metrics

    At present, group animation path in film scenes is generally set manually, existing the problems such as discontinuous path, low animation efficiency, etc. To address the issue, a new algorithm of global path planning for film group animation, which incorporates the Bezier curve and group density information into path editing and optimization based on the jump-point search (JPS) algorithm has been proposed. Firstly, the JPS algorithm is used to generate the group motion path automatically, so that editable path nodes can be used as the control points of the Bezier curve to adjust the path for solving the problems of fold lines, large deflection angles and lack of smoothness in the path. Then, the group density information is set on the path nodes, and the speed and moving direction of the agents are adjusted according to the density information to solve the problems of group motion blockage and individual collision. The experimental results show that the new algorithm is feasible in that it performs well in simulating large-scale group motion and that it improves significantly production efficiency of group animation motion path, especially that in complex film scenes.

    Sentiment analysis of Chinese movie reviews based on deep learning
    ZHOU Jingyi, GUO Yan, DING Youdong
    2018, 24(5):  703-712.  doi:10.12066/j.issn.1007-2861.2075
    Asbtract ( 2770 )   HTML ( 44)   PDF (1030KB) ( 7558 )  
    Figures and Tables | References | Related Articles | Metrics

    With the rise of social networks, more people choose to express their opinions on the internet, which allows film and television investors to collect the audience's feedback more easily. The watercress movie review is just one such platform through which investors are able to know the viewers' taste and preference, and thereby to make better decision in investing the television and film industry. A large amount of data analysis must be done by means of computer technology. Sentiment analysis is a direction of natural language processing (NLP). Sentiment analysis, also known as emotional tendency analysis, is one aiming to analyze the positive or negative aspects of text description. In order to improve the accuracy of the film's sentiment classification, multiple sets of contrast experiments are set to select the optimal parameters, and the Chinese character vectors and the word vectors are compared as the input matrix, in the bidirectional long short-term memory (Bi-LSTM) model and the convolutional neural network (CNN). A Bagging algorithm with CNN model as weak classifier is proposed. Multiple CNN models are trained to determine the final classification results by voting method. The integrated method reduces the deviation caused by a single model. The accuracy of a single Bi-LSTM model has increased by 5.10%, which is 1.34% higher than that of a single CNN model.

    Research Articles
    Bifurcation and chaos of axially moving viscoelastic beam constituted by standard linear solid model
    LI Yi, YAN Qiaoyun, DING Hu, CHEN Liqun
    2018, 24(5):  713-720.  doi:10.12066/j.issn.1007-2861.1870
    Asbtract ( 961 )   HTML ( 15)   PDF (1918KB) ( 282 )  
    Figures and Tables | References | Related Articles | Metrics

    The nonlinear dynamic behavior of an axially moving viscoelastic beam under parametric excitation is investigated. A standard linear solid model is used in the constitutive relation. Newton's second law is applied to derive a nonlinear integral-partial-differential governing equation of the beam. The fourth-order Galerkin truncation method is applied to truncate the governing equation into a set of ordinary differential equations solved with the fourth-order Runge-Kutta method. Based on the diagrams of time history, phase, Poincaré map and frequency analysis, the dynamical behavior is identified. The investigation is focused on the effects of the standard linear solid model's viscoelasticity on the nonlinear dynamic behavior. Numerical simulations show that vibration of an axially accelerating viscoelastic beam is sensitive to all parameters of the standard linear solid model.

    Best constants for two classes of Hausdorff operators on the Heisenberg group
    GUO Jiuhua
    2018, 24(5):  721-729.  doi:10.12066/j.issn.1007-2861.1868
    Asbtract ( 916 )   HTML ( 14)   PDF (495KB) ( 175 )  
    References | Related Articles | Metrics

    The best estimates for two classes of Hausdorff operators on the Heisenberg group was considered. The best constants for the two classes of Hausdorff operators on central Morrey spaces and Morrey spaces were obtained. As applications, the best estimates for the two classes of Hausdorff operators on central BMO (CBMO) spaces were calculated.

    Application of data provenance in multi-version documents retrieval
    CHEN Yue, DONG Hongbin, TAN Chengyu, LIANG Yiwen
    2018, 24(5):  730-744.  doi:10.12066/j.issn.1007-2861.1843
    Asbtract ( 772 )   HTML ( 11)   PDF (5445KB) ( 158 )  
    Figures and Tables | References | Related Articles | Metrics

    As the big data era emerges, the number of document versions is rapidly growing to make document retrieval difficult. Related studies show that provenance information is an important cue in helping users find needed documents. Information retrieval researches based on data provenance often capture files events that cannot describe particular relationship between documents, and therefore are not useful enough for re-finding documents. This paper presents a provenance model based on PROV at the content level, and constructs a specific vocabulary for multi-version documents retrieval. Furthermore, a low-level mode is described with resource description framework (RDF), and the high-level is formed based on query of the former. Finally, to give users a more accessible way to evaluate information, a visualization method of the provenance information is proposed. The results show that the model provides users with more valuable cues by using provenance information to expand retrieval results, and help them find target document quickly and improve efficiency.

    MPPT control of a incremental conductance method with optimized variable step size
    HU Kewei, RAN Feng, QIN Jiaqi, JI Yuan
    2018, 24(5):  745-754.  doi:10.12066/j.issn.1007-2861.1854
    Asbtract ( 1416 )   HTML ( 27)   PDF (13784KB) ( 288 )  
    Figures and Tables | References | Related Articles | Metrics

    As output characteristics of photovoltaic cells change with the change of environment, reasonable maximum power point tracking (MPPT) is a key in improving efficiency of a photovoltaic system. However, there is an inherent contradiction between tracking speed and steady-state accuracy in a fixed step algorithm. The traditional variable step algorithm lacks flexibility, and has a slow start speed, limiting tracking quality of MPPT. An improved variable step incremental conductance method is proposed to solve this problem. The algorithm can select a reasonable proportion coefficient for tracking according to different work locations to overcome problems in the traditional variable step size algorithm. Simulation with Matlab/Simulink shows feasibility of the algorithm under varying illumination conditions.

    Detection and correction of dead pixels in an image
    HU Yunsheng, HU Yueli, YAN Ming, CHENG Wenlong, WANG Quan
    2018, 24(5):  755-762.  doi:10.12066/j.issn.1007-2861.1853
    Asbtract ( 2290 )   HTML ( 77)   PDF (8648KB) ( 752 )  
    Figures and Tables | References | Related Articles | Metrics

    Focusing on image dead pixels due to various factors in the manufacturing process of the imaging sensor and in applications, this paper proposes an algorithm to detect and correct dead pixels. The algorithm judges whether certain dead pixels locate at edges, and if so, calibrates them so that edges are preserved. Experimental results show effectiveness of the algorithm in removing dead pixels. Compared with other algorithms, the algorithm has stronger edge preserving ability, better processing performance and less detection errors.

    Image segmentation for painting characters on billet surface
    HUANG Chunhui, ZHAO Qijie, KE Zhennan
    2018, 24(5):  763-772.  doi:10.12066/j.issn.1007-2861.1862
    Asbtract ( 960 )   HTML ( 15)   PDF (19271KB) ( 125 )  
    Figures and Tables | References | Related Articles | Metrics

    To obtain better segmentation of painting character images on a billet surface, a algorithm is proposed to improve picture quality by image enhancement and filtering, and solve problems in the character image segmentation. An automatic adaptive threshold segmentation algorithm is developed to locate character strings. The algorithm is based on morphology, connected component analysis and twice division to solve the problem of adhesion and fracture present in the image. A filter with a character protection mask is proposed to deal with thin line noises. Experimental results show that the proposed algorithms have good performance for images with character adhesion and fractures, and achieve correct segmentation rates of 99.6% and 98.3%, respectively. The algorithms are also effective to solve the problems of thin line noises, achieving a correct segmentation rate of 97.9%.

    Preparation and properties of PU/HSP composite foams
    LIU Li, JIANG Longsun, HAO Xinmin, CHEN Xiao, HUANG Jie
    2018, 24(5):  773-781.  doi:10.12066/j.issn.1007-2861.1967
    Asbtract ( 1062 )   HTML ( 15)   PDF (9558KB) ( 104 )  
    Figures and Tables | References | Related Articles | Metrics

    Polyurethane (PU) foam reinforced by free-rising hemp stem powder (HSP) is prepared with polyols and polyaryl polymethylene isocyanate (PAPI). Water is used as a blowing agent. Effects of the HSP on the physical properties and bacteria inhibitive rate are studied. The results shows that rebound resilience, tensile strength and elongation at break increase with the increasing HSP content within a certain range. The 75% compression set and bacteria inhibitive increase with the HSP content. The composite foams with 15 phr HSP shows the optimum overall properties with density and the 75% compression set are less than 50 kg/m3 and 5%, respectively. Hardness and tensile strength reach 256.1 N and 124 kPa, and rebound resilience around 50%.

    Preparation and properties of EVA/HSP composite foams
    LIU Li, ZHU Huijie, TU Duoxiang, HAO Xinmin, HUANG Jie
    2018, 24(5):  782-790.  doi:10.12066/j.issn.1007-2861.1956
    Asbtract ( 1113 )   HTML ( 15)   PDF (11226KB) ( 84 )  
    Figures and Tables | References | Related Articles | Metrics

    The composite foams consisting of ethylene-vinyl acetate (EVA) and hemp stem powder (HSP) are prepared by melt mixing and compressing molding. Morphology of the samples is evaluated with scanning electron microscopy (SEM), and rheological properties evaluated with Haake rheometer. The effects of HSP content on quick drying and mechanical properties are studied, and the mechanism is explored. When the HSP content is 10$\sim$20 phr, the materials have the best overall performance because a proper amount of HSP can improve cell morphology, strengthen, toughen the matrix, and give the materials quick drying performance.

    Design and realization of the cloud computing platform for VOCs monitoring and early warning
    ZHANG Tao, JIANG Xiaomeng, JIAO Zheng
    2018, 24(5):  791-799.  doi:10.12066/j.issn.1007-2861.1860
    Asbtract ( 1079 )   HTML ( 14)   PDF (12216KB) ( 92 )  
    Figures and Tables | References | Related Articles | Metrics

    Monitoring and early warning of volatile organic compounds (VOCs) belongs to complicated system engineering. It is a research hotspot and a challenge in environmental science. In this paper, by analyzing shortcomings of existing mechanisms for monitoring and early warning, a VOCs monitoring and early warning cloud platform based on cloud computing is designed. The overall architecture of the system is introduced. The system components and main modules of the platform construction are illustrated. On this basis, key technologies and advantages of the platform are analyzed, showing its value in applications.

    Inhibition of endogenous H$_\textbf{2}$S concentrations in human hepatocyte L02 cells by 1-chloro-2-hydroxy-3-butene
    ZHANG Fang, ZHANG Xinyu
    2018, 24(5):  800-806.  doi:10.12066/j.issn.1007-2861.1825
    Asbtract ( 750 )   HTML ( 11)   PDF (3317KB) ( 130 )  
    Figures and Tables | References | Related Articles | Metrics

    1-chloro-2-hydroxy-3-butene (CHB) is a metabolite of 1,3-butadiene, an environmental pollutant. CHB is cytotoxic, genotoxic, and mutagenic. H$_2$S is the third gasotransmitter following NO and CO, and is involved in a variety of physiological activities. Abnormal metabolism of H$_2$S is associated with many diseases. In the present study, the effect of CHB on the concentrations of endogenous H$_2$S in human hepatocyte L02 cells was investigated. It was observed that CHB could cause significant decreases in the concentrations of endogenous H$_2$S in the cells and the effect depended on CHB concentration. Thus, inhibition of endogenous H$_2$S production in cells by CHB may be a mechanism for toxicity of CHB, which can contribute to a better understanding of toxicity of 1,3-butadiene.

    Robust optimization of production quantity decisions with uncertainty in fuel price and electric power demand
    WANG Jian, WANG Ting, LIU Huixia
    2018, 24(5):  807-818.  doi:10.12066/j.issn.1007-2861.1850
    Asbtract ( 1037 )   HTML ( 11)   PDF (6765KB) ( 77 )  
    Figures and Tables | References | Related Articles | Metrics

    A robust optimization model is established for decisions of electric power production quantity. This model examines impact on the production decisions from two important variables, i.e., fuel price and electric power demand. A new algorithm is developed with SeDuMi in MATLAB. Several suggestions are given based on data analysis. The total production cost of electric power may increase with an increasing uncertainty range of these two variables. If the disturbance directions of the uncertainties in these two variables change in the same direction, the total production cost increases linearly with the increasing uncertainty range. If the disturbance directions of the uncertainties in these two variables change inconsistently, the total production cost increases with increasing volatility. The results show that the changing range of these two variables can also affect the significant level of the relationship between these two variables and the production decisions. This study supports the electric power production quantity decisions under uncertain fuel price and electric power demand.

    Influencing factors of enterprise innovation capacity based on group DEMATEL decision approach of intuitionistic fuzzy preference
    BAO Yuhang, LI Baojun, YU Liying
    2018, 24(5):  819-828.  doi:10.12066/j.issn.1007-2861.1872
    Asbtract ( 1096 )   HTML ( 17)   PDF (2561KB) ( 115 )  
    Figures and Tables | References | Related Articles | Metrics

    This paper establishes an enterprise innovation capacity influencing factor index system from the perspective of innovative ecosystem. The system describes enterprise innovation influencing factors with 15 indicators in 3 dimensions, namely, basic organizing capacity, coordination control capacity, and co-evolution capacity. A DEMATEL decision approach is then developed based on intuitionistic fuzzy preference information, an approach that can be used to get an inner-relationship among the influence factors. Finally, taking an enterprise F as an example, feasibility of the evaluation model is verified.

    Mechanism of patent sharing with government-production-university-institute-application integration: an analysis based on the research of materials genome engineering
    LIU Yan, FAN Xiaojun
    2018, 24(5):  829-840.  doi:10.12066/j.issn.1007-2861.1910
    Asbtract ( 1206 )   HTML ( 14)   PDF (5079KB) ( 109 )  
    Figures and Tables | References | Related Articles | Metrics

    Research of materials genome engineering is a continuously operating project with an important impact on the national economy. Protection of intellectual property is the guarantee of the project's success. Based on the motivation of interest, this paper establishes a game model to analyze variation of the revue under the patent license within and outside the contract. In view of the perspective of sustainable development, a concept of patent sharing is proposed to comply with intellectual property protection of materials genome engineering. Based on a theoretical analysis, the paper proposes a construction strategy of patent sharing mechanism of materials genome engineering with integration of government, production, university, institute, and application.

    Two-stage identification of structural damage based on incomplete static data
    YANG Wanfeng, YANG Xiao
    2018, 24(5):  841-852.  doi:10.12066/j.issn.1007-2861.1844
    Asbtract ( 998 )   HTML ( 13)   PDF (14042KB) ( 195 )  
    Figures and Tables | References | Related Articles | Metrics

    In structural damage identification, traditional methods often result in wrong damage locations and large error in the degree of damages due to limited measurements and too many parameters. Applications of these methods are limited for complex structures. A two-stage identification method for structural damage based on incomplete static data is presented in this paper. The method combines damage localization index based on static strain energy and damage degree computation based on a finite element condensation technique. Possible damage locations of the structure are located with the damage localization index. The degrees of damage are determined by solving the damage state equation with a simulated annealing algorithm. To verify availability and reliability of the proposed method, damaged identifications of a 5 element statically indeterminate, a 13 element statically determinate, and a 10 element statically indeterminate planar trusses are simulated numerically. The results show that the proposed method can effectively identify locations of structural damage and the degrees of the damage. It is robust against measurement noise.