Because of the limit of angle of view(AOV) of IR imaging seeker during the approach of missile and target, the detector can only get the partial image sequence of aircraft nose after "lose point". Recognizin...Because of the limit of angle of view(AOV) of IR imaging seeker during the approach of missile and target, the detector can only get the partial image sequence of aircraft nose after "lose point". Recognizing the axis direction on the basis of partial IR image sequence is a key issue of the advanced IR imaging guide air-to-air missile faced. In this paper, a recognition method was proposed based on the morphological skeleton and modified Hough transform, and this method can recognize correctly the axis direction of aircraft nose in different poses during missile-target encounter. Firstly, the morphological skeleton transform was used for the extraction of skeleton features. Secondly, the modified Hough transform was used for the straight-lines detection. Finally, According to the relations between aircraft nose and axis and invariant of nose features in high-speed IR image sequence, the axis direction can be detected and calculated. Experimental results indicate that the method is feasible and effective, and the precision of axis direction recognized can meet the requirement of accurate burst control of GIF fuze.展开更多
三维点云数据建模是三维重建的核心技术,广泛应用于医学、工业检测及文化遗产保护等领域。文章系统梳理了点云数据的结构与获取方法(如激光扫描、多视图成像),重点分析了基于隐式表示法的表面重建算法。该类算法通过隐式函数描述三维表...三维点云数据建模是三维重建的核心技术,广泛应用于医学、工业检测及文化遗产保护等领域。文章系统梳理了点云数据的结构与获取方法(如激光扫描、多视图成像),重点分析了基于隐式表示法的表面重建算法。该类算法通过隐式函数描述三维表面,擅长处理复杂的拓扑结构,但在面对噪声、稀疏数据及复杂场景时存在局限。为解决这些问题,文章提出了一种基于总变差(Total Variation,TV)正则化的改进算法,利用TV项抑制噪声并保持边界锐利性,结合交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)优化目标函数,显著提升了重建精度与鲁棒性。实验表明,改进算法在噪声环境或数据缺失场景下,能有效消除伪影并保留细节,重建质量优于传统方法。展开更多
拟直流最优潮流(quasi direct current optimal power flow,QDCOPF)模型将复杂的非线性规划问题线性化处理,使得问题模型结构简单,求解速度快,但是计算精度偏低。基于此,提出一种基于改进内点半定规划(modified semidefinite programmin...拟直流最优潮流(quasi direct current optimal power flow,QDCOPF)模型将复杂的非线性规划问题线性化处理,使得问题模型结构简单,求解速度快,但是计算精度偏低。基于此,提出一种基于改进内点半定规划(modified semidefinite programming,MSDP)算法的QDCOPF。一方面,该算法采用半定规划(SDP)算法求解QDCOPF模型,充分利用了SDP鲁棒性强,对初值不敏感和可以收敛到全局最优解的优良性质;另一方面,对映射到SDP模型中的变量进行优化重组,将具有代数关系的变量组成一个变量组,形成一个程序运行子块,加强模型块内部变量联系的同时,减小模型块间变量的耦合。对IEEE 30、118、300节点系统的仿真测试结果表明,基于MSDP的QDCOPF算法比传统QDCOPF的求解算法具有更高的计算精度和计算效率,从而验证了所提算法的可行性和高效性。展开更多
文摘Because of the limit of angle of view(AOV) of IR imaging seeker during the approach of missile and target, the detector can only get the partial image sequence of aircraft nose after "lose point". Recognizing the axis direction on the basis of partial IR image sequence is a key issue of the advanced IR imaging guide air-to-air missile faced. In this paper, a recognition method was proposed based on the morphological skeleton and modified Hough transform, and this method can recognize correctly the axis direction of aircraft nose in different poses during missile-target encounter. Firstly, the morphological skeleton transform was used for the extraction of skeleton features. Secondly, the modified Hough transform was used for the straight-lines detection. Finally, According to the relations between aircraft nose and axis and invariant of nose features in high-speed IR image sequence, the axis direction can be detected and calculated. Experimental results indicate that the method is feasible and effective, and the precision of axis direction recognized can meet the requirement of accurate burst control of GIF fuze.
文摘三维点云数据建模是三维重建的核心技术,广泛应用于医学、工业检测及文化遗产保护等领域。文章系统梳理了点云数据的结构与获取方法(如激光扫描、多视图成像),重点分析了基于隐式表示法的表面重建算法。该类算法通过隐式函数描述三维表面,擅长处理复杂的拓扑结构,但在面对噪声、稀疏数据及复杂场景时存在局限。为解决这些问题,文章提出了一种基于总变差(Total Variation,TV)正则化的改进算法,利用TV项抑制噪声并保持边界锐利性,结合交替方向乘子法(Alternating Direction Method of Multipliers,ADMM)优化目标函数,显著提升了重建精度与鲁棒性。实验表明,改进算法在噪声环境或数据缺失场景下,能有效消除伪影并保留细节,重建质量优于传统方法。
文摘拟直流最优潮流(quasi direct current optimal power flow,QDCOPF)模型将复杂的非线性规划问题线性化处理,使得问题模型结构简单,求解速度快,但是计算精度偏低。基于此,提出一种基于改进内点半定规划(modified semidefinite programming,MSDP)算法的QDCOPF。一方面,该算法采用半定规划(SDP)算法求解QDCOPF模型,充分利用了SDP鲁棒性强,对初值不敏感和可以收敛到全局最优解的优良性质;另一方面,对映射到SDP模型中的变量进行优化重组,将具有代数关系的变量组成一个变量组,形成一个程序运行子块,加强模型块内部变量联系的同时,减小模型块间变量的耦合。对IEEE 30、118、300节点系统的仿真测试结果表明,基于MSDP的QDCOPF算法比传统QDCOPF的求解算法具有更高的计算精度和计算效率,从而验证了所提算法的可行性和高效性。
