共查询到19条相似文献,搜索用时 161 毫秒
1.
基于轮廓特征的图像配准 总被引:1,自引:0,他引:1
提出了一种新的基于轮廓特征的图像配准算法,首先提取图像轮廓并计算每个轮廓点的法向角,然后对轮廓点法向角进行直方图统计,通过对两幅图像的轮廓点法向角直方图进行圆周相关计算便可快速估计出两幅图像所存在的旋转角度.由于旋转角度参数的求出通常可以大大简化配准变换模型中其他参数的估计,因而这种方式可以实现快速配准.这种利用轮廓点法向角来估计旋转角度的方式具有旋转、平移和尺度不变性,并对轮廓缺失以及存在噪声的情况均具有很高的鲁棒性,可广泛适用于存在闭合轮廓与开轮廓的各种情况.本文所提出的配准算法已成功应用于PCB缺陷检测中实现了参考图像轮廓和待检测图像轮廓的快速精确配准. 相似文献
2.
3.
4.
本文提出了一种多重约束下由粗到精的多源图像自适应子像素级配准算法.该算法采用影像特征点作为匹配基元,利用具有不同精度等级的组合判据法、整体松弛法、最小二乘法实现由粗到精的匹配,同时在匹配过程中加入了多重约束,如定位点控制约束、交叉匹配约束、连续控制约束,以保证获取的配准控制点的可靠性和剔除粗差点.此外,该算法利用配准控制点自适应地构建整个图像的三角网,最后依据改进的三角形填充算法对目标图像进行逐像点纠正.对同源和非同源的遥感图像的实验证明,SPOT4全色图像(10m/pixel)和SPOT5多光谱图像(10m/pixel)的配准精度分别达到6~7m和5~6m. 相似文献
5.
具有SIFT描述的Harris角点多源图像配准 总被引:3,自引:0,他引:3
多源传感器成像原理的差异给图像配准带来了很大困难,本文针对红外与可见光图像配准提出了一种具有SIFT描述特征的Harris角点多源图像配准算法。首先建立多尺度空间,以多尺度空间检测尺度不变的Harris角点作为特征点;然后通过改进SIFT对特征点的描述方法,采用圆环结构算子对Harris角点进行类SIFT的特征描述;最后利用双向最近邻方法进行匹配,通过最小二乘法实现图像的配准。实验证实了算法配准的精确性、快速性和稳定性,具有较好的配准效果。 相似文献
6.
基于仿射不变SIFT特征的SAR图像配准 总被引:2,自引:2,他引:0
针对SAR(Synthetic Aperture Radar)图像全自动配准问题,本文提出一种基于仿射不变SIFT(Scale Invariant Feature Transform)特征的精确配准方法.该方法首先对传统SIFT方法改进构建具有仿射不变性的SIFT描述子,并利用该描述子对提取的控制点进行粗匹配,然后由粗匹配点对的尺度比和方位差及其邻域的灰度相似性构建新的相似矩阵,最后利用SVD(Singular Value Decomposition)方法确定精确匹配点对,求出变换参数从而实现图像的精确配准.实验结果表明该方法优于传统的SIFT方法和SIFT+SVD方法并且可以达到亚像素的配准精度. 相似文献
7.
8.
目的 为了提高锂电池丝印图像配准精度,从而解决产品质量检测中的漏检和误报问题,研究点特征提取算法在锂电池丝印图像配准中的应用.方法 对基于点特征的锂电池丝印图像配准进行综述,首先概述点特征提取算法的发展历程,然后着重围绕Harris,SIFT,SURF,ORB和AKAZE等5种经典的点特征提取算法进行分析,并介绍近几年的提升算法,最后对锂电池丝印图像进行配准测试,利用几种测评技术对实验效果进行分析,总结不同点特征提取算法在锂电池丝印图像配准中的优缺点和适用性.结果 实验结果表明,AKAZE算法提取的特征点具有较高的重复率和匹配准确率,经过配准后的定位误差也都控制在1个像素以内,但是该算法的尺度不变性较差.结论 相较于前4种算法,AKAZE算法具有较高的可靠性和稳定性,能够满足锂电池丝印图像配准的实时性和高效性需求. 相似文献
9.
目的为了解决当前图像配准算法因利用l1距离或l2距离相似度测量手段来完成图像特征点匹配,使其忽略了相位信息,难以有效消除高斯噪声的影响,使其配准精度与效率不佳不足的问题。方法提出最优相似度距离耦合角度径向变换的抗噪图像配准算法。首先引入角度径向变换,以降低算法复杂度,快速提取图像的特征点。然后联合图像的幅度和相位信息,基于欧式距离测度,定义最优相似度距离测量模型,通过求解其全局最小值,对特征点完成匹配,提高算法的抗噪性能。最后将图像分割为内点与外点,择取6个内点,通过计算其变换矩的几何配准误差,改进随机样本一致策略,对匹配进行提纯,消除误配。结果仿真实验结果显示,与当前基于l1距离或l2距离相似度测量的图像配准技术相比,该算法具有更强的抗高斯噪声性能和更高的匹配精度,且算法时耗最短。结论所提算法能够精确完成图像特征配准。 相似文献
10.
由于卫星图像来自于不同的传感器、由不同的视角和光谱、在不同的时间获得,图像间存在较大差异。为了有效配准图像,提出一种"先粗后精"的配准算法,首先采用Fourier-Mellin变换算法实施快速的粗配准,然后采用以修正的结构相似度为测度的优化算法实施精确配准。对于真实的卫星图像配准,由于没有准确的衡量标准,很难给出定量的评估结果。本文提出一种新的配准评估方法?匹配曲线特征评估法,以匹配曲线的峰度、峰偏、峰值以及峰值间均方根误差(RMSE)为定量评估指标,以峰值间RMSE最小为准则自动调整配准参数。结果表明,"先粗后精"的配准算法能够实现相当精确的配准;匹配曲线特征评估法不仅能够从曲线的光滑度、尖锐度等特性直观描述配准性能,并能由曲线的特征指标定量评估配准效果,而且还能自动调整配准参数,使配准更加精确。 相似文献
11.
12.
针对目前图像拼接中计算量大、精度低两大难题,本算法应用高精度伺服控制平台、边缘信息阈值分割和模板匹配图像处理方法,实现了PCB图像的快速拼接。由于高精度伺服控制平台的引入,使得图像重合区域的参数获取变得容易。实验表明:应用伺服控制平台以后,提高了算法的实时性,达到较高的精度。 相似文献
13.
基于虚拟点的可见光和SAR图像配准研究 总被引:1,自引:1,他引:0
本文以机场场景下的可见光和SAR图像为研究对象,提出了一种基于虚拟点特征的可见光和SAR图像配准方法.该方法以虚拟点特征和控制点匹配技术为基础,处理具有全局仿射几何失真的异源图像配准问题.首先根据两类图像的特点,使用Canny算子和一种兴趣算子提取两幅图像中的共有特征一直线特征,然后在直线特征的基础上拟合虚拟点特征,采用基于特征一致的粗配准和基于虚拟点特征的精确配准相结合的方法,对两幅图像实现由粗到精的自动配准,实验结果表明,本文方法可行且能取得较高的配准精度. 相似文献
14.
为改进SAR图像匹配的稳健性和实时性,提出一种基于小波变换的等价图割SAR图像配准方法.该方法首先利用小波变换对图像进行分解,在低频子图像下构造等价图割,克服相干斑噪声干扰,避免NP困难,解决映射函数选取问题,从图像中分割出精确目标.其次利用尺度不变特征变换(SIFT)方法实现目标的特征匹配,降低搜索空间特征点描述,提高实时性.最后通过匹配关系找到变换参数,实现图像精确配准.实验结果表明,该方法能快速而精确地实现SAR图像配准. 相似文献
15.
Alireza Ahmadian Nasim Dadashi Serej Saeed Karimifard Parastoo Farnia 《International journal of imaging systems and technology》2013,23(4):294-303
Estimation of soft tissue deformation occurring during image‐guided surgery using an easily implemented and accurate method is necessary. Using a stereo camera, this study focuses on two efficient methods for estimating soft tissue deformation. Two methods were proposed to overcome limitations associated with the typical methods used for estimating soft tissue deformation, such as dependence on accuracy of the operator and indentation of skin. The first method is based on Triclops SDK, and the second method is based on projecting a pattern to acquire P‐Lands (Projected Landmarks). Based on the proposed methods, surface information is acquired in the form of point clouds of surface point coordinates to the submillimeter accuracy. The reconstructed predeformation three‐dimensional (3D) point cloud obtained for each method is registered with a modified iterative closest point algorithm to a postdeformation 3D point cloud obtained from the same region of interest. Results were compared with an MRI–MRI registration method as a control. Results are provided as RMS differences between the initial and final coordinates of corresponding points. The average RMS difference for the typical method is 3.53 mm, that for the Triclops SDK method is 2.32 mm, and that for the P‐Lands projection method is 2.06 mm. The MRI‐MRI registration had an average RMS difference of 1.12 mm. Using MRI‐MRI registration as the gold standard, the average error obtained for the typical method was 2.41 mm, that for the first method was 1.2 mm, and that for the second method was 0.94 mm. © 2013 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 23, 294–303, 2013 相似文献
16.
17.
18.
The design and testing of an efficient method for measurement of the location of corners of binary regions in digital images is described. From a single perspective image, the image coordinates of corner points are obtained from moments and intersection points accumulated within specified windows. Corner-point coordinates may then be used in further processing such as inverse photogrammetric solutions to determine 3D position. A video-processing board based on a TMS 320C25 DSP chip has been developed to process the image windows. Together with an 80386-based single-board computer, object corner location is achieved at the standard RS-170 television field rate of 60 Hz 相似文献
19.
Peng Gui Wing-Kuen Ling Dengyi Zhang Yan Xiang Dangguo Shao Lei Ma 《International journal of imaging systems and technology》2019,29(4):701-710
Image registration is the process of overlaying images of the same scene taken at different times by different sensors from different viewpoints. The cross-cumulative residual entropy (CCRE)-based medical image registration could achieve a high precision and a strong robustness performance. However, the optimization problem formulated by CCRE consists of some local extrema, especially for noise images. In order to address these difficulties, this article proposes a new optimization algorithm named hybrid differential search algorithm (HDSA) to optimize CCRE. As HDSA consists of simple control parameters, it is independent of the initial searching point. In addition, HDSA ameliorated the search method and the iterative conditions. As a result, the optimization process is more stable and efficient. Image registration experiments of HDSA are performed and compared with the conventional differential search algorithm (DSA) and adaptive differential evolution with optional external archive (JADE). The results show that HDSA does not only overcome the difficulties of sticking in the local extrema but also enhances the precision of registration. It is effective, robust, and fast for both the single-mode rigid medical image registration and the multispectral-mode rigid medical image registration. 相似文献