针对UAV(Unmanned Aerial Vehicle)侦察目标识别中的实时性和鲁棒性的要求,提出一种基于SURF(Speeded Up Robust Features)的快速目标识别算法。对UAV侦察图像进行预处理,采用不变矩构造遗传算法的适应度函数,利用遗传算法的全局搜索能...针对UAV(Unmanned Aerial Vehicle)侦察目标识别中的实时性和鲁棒性的要求,提出一种基于SURF(Speeded Up Robust Features)的快速目标识别算法。对UAV侦察图像进行预处理,采用不变矩构造遗传算法的适应度函数,利用遗传算法的全局搜索能力快速地提取可能包含目标的RO(IRegion Of Interesting)区域。在ROI区域和模板图像中提取SURF特征点,采用最近邻的匹配算法搜索匹配对,从而精确确定目标的位置。仿真结果显示,该算法可以明显地提高目标识别的实时性并具有相当的鲁棒性。展开更多
利用草图进行图像检索的难点在于对不同尺度、位置、旋转及形变图像的有效检索。为了更准确地识别并检索不同尺度、位置和旋转的图像,提出一种基于草图局部几何不变矩的图像检索方法(SBIRULGMI)。首先,利用图像的几何特征分别确定各图...利用草图进行图像检索的难点在于对不同尺度、位置、旋转及形变图像的有效检索。为了更准确地识别并检索不同尺度、位置和旋转的图像,提出一种基于草图局部几何不变矩的图像检索方法(SBIRULGMI)。首先,利用图像的几何特征分别确定各图像的坐标系;然后,在生成的坐标系中对图像进行平均分块并计算各块的几何不变矩作为特征向量;接着,用改进的欧氏距离计算目标图像与数据库图像的相似度;最后,采用蚁群(ACO)算法对按照相似度排序后的检索结果进行优化。所提方法在MPEG-7 shape1 part B图像数据库的检索识别准确率比形状上下文(SC)、边缘分布直方图(EOH)、局部线性高波特征(GALIF)及Mind Finder方法平均提高了17个百分点。实验结果表明该方法对不同平移、缩放和翻转的图像有较好的识别效果,对图像一定程度的旋转和形变具有更好的鲁棒性。展开更多
We investigate a generalized form of the Boussinesq equation, relevant for nerve pulse propagation in biological membranes. The generalized conditional symmetry (GCS) method is applied in order to obtain the conditi...We investigate a generalized form of the Boussinesq equation, relevant for nerve pulse propagation in biological membranes. The generalized conditional symmetry (GCS) method is applied in order to obtain the conditions that enable the equation to admit a special class of second-order GCSs. For the case of quadratic nonlinearities, we outline a new class of invariant solutions.展开更多
The cyclic transformation behaviors of polycrystalline super-elastic NiTi shape memory alloys (SMAs)under multiaxial loading paths with different angles between axial and torsional loading orientations were experiment...The cyclic transformation behaviors of polycrystalline super-elastic NiTi shape memory alloys (SMAs)under multiaxial loading paths with different angles between axial and torsional loading orientations were experimentally investigated.The experimental results showed that the start stresses of forward and reverse transformations decreased with the increase'in the number of cycles and exhibit obvious anisotropic evolutions.The start stresses of forward and reverse transformations in the tensile and torsional directions did not satisfy the yon Mises criterion.The shape of transformation surface during the forward and reverse transformations evolved with the increase in the number of cycles.Then,new cyclic anisotropic transformation surfaces were established by introducing an anisotropic tensor into the von Mises equivalent stress based on a typical transformation criterion related to J2 and J3.Moreover,the evolution equations of material parameters used in the proposed transformation surfaces were established to describe the subsequent evolutions of transformation surfaces.Finally,the start stresses of forward and reverse transformations predicted using the proposed transformation surfaces were compared with the experimental results.It shows that the proposed transformation surfaces can reasonably describe the start stresses of forward and reverse transformations,which are helpful for establishing a three-dimensional cyclic constitutive model to describe the cyclic transformation behaviors of super-elastic NiTi SMAs.展开更多
文摘针对UAV(Unmanned Aerial Vehicle)侦察目标识别中的实时性和鲁棒性的要求,提出一种基于SURF(Speeded Up Robust Features)的快速目标识别算法。对UAV侦察图像进行预处理,采用不变矩构造遗传算法的适应度函数,利用遗传算法的全局搜索能力快速地提取可能包含目标的RO(IRegion Of Interesting)区域。在ROI区域和模板图像中提取SURF特征点,采用最近邻的匹配算法搜索匹配对,从而精确确定目标的位置。仿真结果显示,该算法可以明显地提高目标识别的实时性并具有相当的鲁棒性。
文摘利用草图进行图像检索的难点在于对不同尺度、位置、旋转及形变图像的有效检索。为了更准确地识别并检索不同尺度、位置和旋转的图像,提出一种基于草图局部几何不变矩的图像检索方法(SBIRULGMI)。首先,利用图像的几何特征分别确定各图像的坐标系;然后,在生成的坐标系中对图像进行平均分块并计算各块的几何不变矩作为特征向量;接着,用改进的欧氏距离计算目标图像与数据库图像的相似度;最后,采用蚁群(ACO)算法对按照相似度排序后的检索结果进行优化。所提方法在MPEG-7 shape1 part B图像数据库的检索识别准确率比形状上下文(SC)、边缘分布直方图(EOH)、局部线性高波特征(GALIF)及Mind Finder方法平均提高了17个百分点。实验结果表明该方法对不同平移、缩放和翻转的图像有较好的识别效果,对图像一定程度的旋转和形变具有更好的鲁棒性。
文摘We investigate a generalized form of the Boussinesq equation, relevant for nerve pulse propagation in biological membranes. The generalized conditional symmetry (GCS) method is applied in order to obtain the conditions that enable the equation to admit a special class of second-order GCSs. For the case of quadratic nonlinearities, we outline a new class of invariant solutions.
基金National Natural Science Foundation of China (1157226511532010),the Excellent Youth Found of Sichuan Province (2017JQ0019),the Open Project of Traction Power State Key Laboratory (TPL1606)and the Exploration Project of Traction Power State Key Laboratory (2017TPL_T04)are acknowledged.
文摘The cyclic transformation behaviors of polycrystalline super-elastic NiTi shape memory alloys (SMAs)under multiaxial loading paths with different angles between axial and torsional loading orientations were experimentally investigated.The experimental results showed that the start stresses of forward and reverse transformations decreased with the increase'in the number of cycles and exhibit obvious anisotropic evolutions.The start stresses of forward and reverse transformations in the tensile and torsional directions did not satisfy the yon Mises criterion.The shape of transformation surface during the forward and reverse transformations evolved with the increase in the number of cycles.Then,new cyclic anisotropic transformation surfaces were established by introducing an anisotropic tensor into the von Mises equivalent stress based on a typical transformation criterion related to J2 and J3.Moreover,the evolution equations of material parameters used in the proposed transformation surfaces were established to describe the subsequent evolutions of transformation surfaces.Finally,the start stresses of forward and reverse transformations predicted using the proposed transformation surfaces were compared with the experimental results.It shows that the proposed transformation surfaces can reasonably describe the start stresses of forward and reverse transformations,which are helpful for establishing a three-dimensional cyclic constitutive model to describe the cyclic transformation behaviors of super-elastic NiTi SMAs.