ID 原文 译文
11634 最后根据匹配结果确定目标在当前帧的位置和尺度。 According to the matching results to determine the location of the target in the current frame and scale. ‭
11635 改进的联合匹配策略在构建相似度矩阵时,不但利用了具有旋转和尺度不变性的SIFT特征向量,并且充分考虑了特征点的空间位置信息,有效提高了特征匹配的准确性。 Improve the joint matching strategy in constructing similarity matrix, by utilizing the rotation and scaling invariance of SIFT feature vector, and fully considering the space of the feature points location information, improve the accuracy of the feature matching.
11636 将这种改进的联合匹配策略成功地引入到SIFT匹配跟踪中,克服了传统SIFT匹配算法用于非刚体目标跟踪时的缺陷。 ‭Will improve the joint match of this strategy successfully introduced to SIFT matching tracking, overcomes the traditional SIFT matching algorithm used for the defects of non-rigid target tracking.
11637 实验结果表明,该算法对目标的非刚性形变、尺度变化以及背景干扰都具有较强的鲁棒性。 ‭The experimental results show that the algorithm of target of non-rigid deformation, dimension change and background interference has strong robustness.
11638 全仿射形变条件下,待配准合成孔径雷达(synthetic aperture radar,SAR)图像与参考SAR图像之间存在各向异性尺度变化,导致传统的点特征图像配准算法难以提取到足够多的匹配特征点进行图像配准。 Full affine deformation conditions, for registration of synthetic aperture radar (synthetic aperture radar, SAR) image and the reference SAR image between anisotropic scale change, lead to some characteristics of the traditional image registration algorithm is difficult to extract enough matching feature points of image registration.
11639 为此,提出了一种基于仿射形变矩阵分解与尺度变化矩阵估计的点特征图像配准算法。 To this end, this paper proposes a matrix decomposition based on affine deformation matrix estimation and scale change point feature image registration algorithm.
11640 该方法首先将仿射形变矩阵分解为图像旋转矩阵、尺度变化矩阵以及常数矩阵的乘积。 This method will first affine deformation matrix decomposition of the image rotation matrix, dimension change matrix and the product of the constant matrix.
11641 而后利用粒子群优化(particle swarm optimization,PSO)算法对尺度变化矩阵中的未知参数进行搜索估计,并根据估计结果对图像进行尺度规范处理,以抑制图像间的各向异性尺度变化。 Then using the particle swarm optimization (particle swarm optimization, PSO) algorithm estimates the unknown parameters in the matrix search to scale change, and according to the estimation results of image processing dimension specification, to suppress the image between the anisotropy of scale change.
11642 在此基础上再利用尺度不变特征转换(scale invariant feature transform,SIFT)算子提取匹配特征点进行配准处理。 On the basis of reuse scale invariant feature transform (scale invariant feature transform, SIFT) operator registration processing extraction and matching feature points.
11643 实验结果表明,与现有方法相比,对于全仿射形变条件下的SAR图像配准,本文所述算法可以提取到更多的匹配特征点,因而具有更好的配准性能。 The experimental results show that, compared with existing methods, for the whole affine deformation under the condition of SAR image registration, this article described algorithm can extract more matching feature points, thus has better matching performance.