ID 原文 译文
11704 研究了一种基于虚拟变偏振理论的红外面目标细节增强方法。 Study a red outside target based on virtual variable polarization theory details enhancement method.
11705 该方法充分挖掘和利用了红外偏振信息的固有特点,利用入射光Stokes矢量和出射光光强之间的描述关系推导出出射光光强和起偏角度的解析关系,并据此虚拟实现了任意起偏角度的出射光光强。 This method fully mining and use of the inherent characteristics of the infrared polarization information, using the incident light Stokes vector and emergent light strong guy parachuting is derived and a description of the relationship between strong and slant Angle analytical relations, and on the basis of virtual achieved arbitrary slant Angle of emergent light is strong.
11706 然后依据面目标信杂比最大准则,利用粒子群算法迭代实现了最优起偏角度的搜索,最终得到增强后的红外面目标图像。 And on the basis of surface target shift-and-correlate than maximum criterion, using the iterative particle swarm algorithm to achieve the optimal slant Angle search, finally obtained the enhanced red outside the target image.
11707 利用实测的长波红外偏振图像数据对算法的可行性和有效性进行了验证。 Using the measured long wave infrared polarization imaging data on the feasibility and effectiveness of this algorithm is verified.
11708 实验结果表明,在灰度对比度、平均梯度、图像熵3种图像质量评价指标下,经此算法增强后的图像质量更高,为后续目标细节识别和攻击点选择等算法的实现奠定了良好的数据基础。 The experiment results show that the gray contrast, average gradient, image entropy under three kinds of image quality evaluation index, after the algorithm to enhance the image quality is higher, details for later target recognition and focal point selection algorithm to realize data laid a good foundation.
11709 采用了一种空间敏感度特征包(spatially-sensitive bags of feature,SS-BOF)来实现合成孔径雷达(synthetic aperture radar,SAR)图像的地物识别。 A spatial sensitivity was used to wrap (spatially - sensitive bags of feature, SS - BOF) for synthetic aperture radar (synthetic aperture radar, SAR) image feature recognition.
11710 首先采用推广的核模糊C-均值方法分割SAR图像,提取SAR图像目标图形; ‭First nuclear fuzzy C - average method is used to SAR image segmentation and extraction of SAR image target graphics;
11711 采用Harris角点检测子提取角点,接着对目标图形进行Delaunay三角剖分; The Harris corner detection to extract the corner, then Delaunay triangle subdivision of target graphics;
11712 采用cotangent weight方法对三角剖分图赋值,进而求得离散化Laplace-Beltrami算子的特征值、特征向量,并计算SS-BOF,进而对地物目标进行识别,其识别方法采用比L1相似准则效果更好的相关系数法; The cotangent weight method of triangle subdivision graph assignment, then the discretization of Laplace - Beltrami operator eigenvalue and eigenvector, and calculate the SS - BOF, and then to ground object target recognition, the recognition method using better than L1 similarity criterion of correlation coefficient method;
11713 最后与热核迹等热核不变量特征以及Hu不变矩特征进行对比。 Finally with thermonuclear fusion invariant characteristics such as trace and Hu moment invariant features were compared.