基于光学GCP图像切片的SAR图像自动匹配方法

地面控制点是进行星载SAR图像几何精校正的重要条件之一.通过光学GCP切片和SAR原始图像进行匹配可以实现控制点的自动提取.提出了一种在图像梯度域,实现光学图像和SAR图像异源匹配的方法.首先根据SAR的轨道参数等信息,实现光学GCP切片和SAR图像尺度和旋转的归一化,解决异源图像间的分辨率和旋转角度差异.然后分别提取SAR图像和光学图像的梯度强度信息,为了消除SAR图像的斑点噪声,采用ROA算法准确提取SAR图像的梯度强度.SAR图像和光学图像的成像机理差别很大,直接在图像域匹配较为困难.但是它们的梯度强度图像具有相似性,为此可以通过在梯度强度图像域,应用归一化互相关匹配算法,实现同名点的自动匹配.最后采用实验证明了方法的可行性.     

Automated segmentation and volumetric analysis of renal cortex,medulla, and pelvis based on non-contrast-enhanced T1- and T2-weighted MR images

Object

The aim of our study was to enable automatic volumetry of the entire kidneys as well as their internal structures (cortex, medulla, and pelvis) from native magnetic resonance imaging (MRI) data sets.

Materials and methods

Segmentation of the entire kidneys and differentiation of their internal structures were performed in 12 healthy volunteers based on non-contrast-enhanced T1- and T2-weighted MR images. Two data sets (each acquired in one breath-hold) were co-registered using a rigid registration algorithm compensating for possible breathing-related displacements. An automatic algorithm based on thresholding and shape detection segmented the kidneys into their compartments and was compared to a manual labeling procedure.

Results

The resulting kidney volumes of the automated segmentation correlated well with those created manually (R ² = 0.96). Average volume errors were determined to be 4.97 ± 4.08 % (entire kidney parenchyma), 7.03 ± 5.56 % (cortex), 12.33 ± 7.35 % (medulla), and 17.57 ± 14.47 % (pelvis). The variation of the kidney volume resulting from the automatic algorithm was found to be 4.76 % based on the measuring of one volunteer with three independent examinations.

Conclusion

The results demonstrate the feasibility of an accurate and repeatable automatic segmentation of the kidneys and their internal structures from non-contrast-enhanced magnetic resonance images.     

Automatic MRI adipose tissue mapping using overlapping mosaics

This paper presents an automatic method of correcting non-uniform RF coil response for the classification of body composition using MR imaging. By linear mosaic modelling, the smoothly but non-linearly varying bias field, which modulates tissue intensities within the image, was corrected. The overlapping between adjacent mosaics ensured consistent segmentation of body fat content and the effectiveness of the technique was validated by both phantom and in vivo experiments. Ten whole body composition data sets, each with 39 trans-axial slices, were acquired. Automatic segmentation results using the proposed technique were compared with those from manual delineations. The automatic segmentation method was found to be highly accurate and the mean percentage error between the two methods was less than 1.5%.     

实数编码遗传模拟退火算法SHEPWM控制技术

针对多电平逆变器SHEPWM(selected harmonic elimination pulse width modulation,SHEPWM)消谐模型的求解问题,提出一种遗传算法和模拟退火算法相融合的新型算法。该算法根据个体适应度值进行自适应交叉和变异操作,采用模拟退火算法进行个体更新,以增加种群的多样性,增强全局寻优能力,避免陷入局部最优,并以实数编码保存个体来提高计算精度。以二极管箝位三电平逆变器SHEPWM为例,给出了全调制度下的开关角度轨迹及较高调制度下的另外两组解,绘制了谐波失真含量(total harmonic distortion,THD)随调制度变化的曲线,并给出了详细的仿真结果。最后通过建立的二极管箝位三电平逆变器实验平台进行了实验验证,仿真和实验结果证明了该算法的正确性和可行性。     

A scale space based algorithm for automated segmentation of single shot tagged MRI of shearing deformation

Object

This study proposes a scale space based algorithm for automated segmentation of single-shot tagged images of modest SNR. Furthermore the algorithm was designed for analysis of discontinuous or shearing types of motion, i.e. segmentation of broken tag patterns.

Materials and methods

The proposed algorithm utilises non-linear scale space for automatic segmentation of single-shot tagged images. The algorithm's ability to automatically segment tagged shearing motion was evaluated in a numerical simulation and in vivo. A typical shearing deformation was simulated in a Shepp-Logan phantom allowing for quantitative evaluation of the algorithm's success rate as a function of both SNR and the amount of deformation. For a qualitative in vivo evaluation tagged images showing deformations in the calf muscles and eye movement in a healthy volunteer were acquired.

Results

Both the numerical simulation and the in vivo tagged data demonstrated the algorithm’s ability for automated segmentation of single-shot tagged MR provided that SNR of the images is above 10 and the amount of deformation does not exceed the tag spacing. The latter constraint can be met by adjusting the tag delay or the tag spacing.

Conclusion

The scale space based algorithm for automatic segmentation of single-shot tagged MR enables the application of tagged MR to complex (shearing) deformation and the processing of datasets with relatively low SNR.     

基于旋转块的BM3D图像去噪改进算法

针对经典BM3D去噪算法中存在的相似块匹配误差较大及对图像细节保护不足的问题,提出了一种基于旋转块的BM3D图像去噪改进算法.新算法首先对参考块进行不同角度的旋转获取旋转块,通过旋转块进行相似块匹配过程;然后使用低秩正则化来代替传统算法中的硬阈值滤波;最后,对结合旋转块匹配与低秩正则化的BM3D算法进行自适应调整,从而改善在均匀图像区域中的去噪效果.实验结果表明,新算法的相似块匹配程度更高,峰值信噪比相较于经典算法平均提升0.5 dB,有效保留图像边缘和纹理细节.     

An improved two‐stage image registration algorithm for super‐resolution

This paper presents an improved two‐stage image registration algorithm for super‐resolution. The algorithm is based on the rotation–translation (RT) model and the coarse‐to‐fine strategy. It first uses the phase correlation algorithm to estimate large‐scale displacements with pixel‐level accuracy after image motion compensation, and then uses the Keren algorithm to obtain high‐accuracy sub‐pixel estimation. Moreover, from the non‐commutative property between rotation and translation in the RT model, synthesis formulae are derived and used to combine the two results together, which could further improve the accuracy without extra computational costs. The algorithm can achieve high‐accuracy sub‐pixel registration for large‐scale displacements, which also has the advantage of good computational efficiency. To illustrate the effectiveness of the algorithm, both simulations and practical super‐resolution reconstruction experiments are performed. © 2014 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

一种新颖的虹膜识别算法

提出一种新颖的虹膜识别算法.该算法利用二维Log-Gabor滤波器对归一化后的虹膜图像进行滤波,滤波后的中高频子带图像被分割为若干子块,提取各子块局部极大值点并对其进行相位粗量化编码来生成虹膜特征码;特征匹配时使用虹膜特征码和干扰掩模来生成汉明距离,以此衡量两个虹膜之间的相似程度.该方法具有平移、放缩和旋转不变性,并且生成的虹膜特征码仅288 B.实验结果和数据分析表明该方法的合理性和可行性.     

Automated registration of brain images using edge and surfacefeatures

We present an automated multimodality (CT & MRI) registration algorithm based on hierarchical feature extraction. Two kinds of shape representations-edge and surface-are extracted hierarchically from different image modalities. The registration then is performed using least-squares matching of the user-specified (but automatically extracted) corresponding features. In our implementation, the 3-D version of the Canny edge detector is employed in the extraction of corresponding edge information. An automatic segmentation algorithm is introduced to extract the corresponding surfaces from the edges efficiently. The geometric matching of those extracted shape features then is performed using the iterative closest-point matching method     

不同分布下红外主动目标特征点配准方法研究

视觉测量中经常需要用到特征匹配来计算位姿信息,但尚无针对红外主动目标设计特征匹配可用的算法,为实现对不同分布红外主动目标的匹配,本文提出了一种通用的两阶段特征点匹配方法。第一阶段为粗配准,首先检测图像特征点集的凸包,获取最外围点;通过构建三角形特征集,并使用马氏距离进行相似三角形计算搜索实现快速粗配准。第二阶段为精匹配,首先通过粗匹配特征计算欧拉角避免匹配结果的180°旋转对称;针对粗配准后的可能存在的特征点缺失问题,采用极线约束精匹配策略,充分利用已匹配特征点的几何信息,有效实现对剩余点的精确匹配。理论分析与实验表明,在13个红外发光点组成的旋转对称点集及非旋转对称点集下,该方法在绝对大小0°~40°的旋转范围内能高效匹配,实验测试极限性能能够达到50°,并对实际场景下特征点的遮挡等情况具有较好的鲁棒性,实验结果验证其适应性与稳定性,具有较高的实用价值。     

DAISY和LBP描述符结合的三维点云粗拼接方法

针对三维点云拼接需借助点云信息几何特征的问题,采用DAISY描述符和LBP描述符结合的方法提取相邻测量站位重叠区域图像特征,解算出相邻测量站位坐标系之间的位置变换矩阵,从而将多测量站位的三维点云数据初步转换至同一坐标系中。首先,介绍了边缘检测和DAISY描述符的构建;然后,通过欧氏距离对相邻图片特征点进行匹配,根据匹配点之间的关系解算出不同站位下的坐标转换关系。实验结果表明,该方法在不使用其他辅助工具的前提下可以较好实现三维点云数据粗拼接,为点云拼接技术在三维重建和逆向工程等领域的应用提供了理论依据。     

高效高精度光照自适应的 ORB 特征匹配算法

针对非均匀光照下ORB图像特征检测算法存在特征点过于聚集、匹配准确率不高等问题,提出了一种高效高精度光照自适应的ORB图像特征匹配算法。利用自适应阈值提取待测图像的oFAST特征点,通过优化的四叉树分解法均匀分配,进一步提高了低照度或高曝光区域特征点的数量,随后,根据汉明距离进行特征匹配,使用改进的RANSAC算法剔除误匹配,提高ORB算法中特征点的匹配准确率。实验结果表明,针对具有明显光照变化的数据集,相较于ORB、MA、Y-ORB及S-ORB算法,本文算法的平均特征分布均匀度提高13.1%,特征提取时间节省26.3%,综合评价指标提升18.5%,可高效完成复杂场景变化下的特征匹配,对目标识别和三维重建等领域具有较强的应用价值。     

基于改进SURF算子的彩色图像配准算法

为了解决传统SURF算法存在的问题,提高彩色图像配准的精度和准确率,提出一种双向邻近匹配的彩色图像配准算法。该算法对传统的SURF描述符进行改进,将图像的色彩信息叠加在只包含灰度信息的传统SURF特征描述符上,组成改进的SURF特征描述符,以增强彩色信息对配准的影响,提高配准的准确率;采用FLANN算法搜索匹配点对,并对匹配点对进行双向邻近匹配,以提高搜索效率和匹配精度;利用改进RANSAC算法剔除匹配错误的特征点对,以进一步优化匹配结果。实验结果表明,该算法能够有效地提高彩色图像配准的精度和准确率,具有较好的鲁棒性和图像变换适应性。     

基于KAZE的自适应模糊图像配准算法

为了提高图像配准算法对于模糊图像的配准性能,提出一种融合非线性尺度空间和空间余弦相似度的自适应模糊图像配准算法。该算法将非线性尺度空间理论应用于图像的局部特征提取,采用KAZE算法提取图像的特征点,以构成M-SURF特征描述符;利用空间余弦对图像特征点进行匹配,并且根据不同的图像特性进行自适应阈值匹配,以得到便于寻求最优变换关系的合理数量的匹配点对;最后采用RANSAC算法滤除误匹配点对,以提升算法精度。实验结果表明,该算法可以有效地提高模糊图像配准的匹配准确率和精度,准确率和精度比KAZE算法最大可以提高25%和7.909像素,具有更好的配准性能。     

量子遗传算法优化的最小二乘支持向量机的风机故障诊断

针对最小二乘支持向量机（LSSVM）在故障诊断过程中的模型参数选择问题,提出了利用全局寻优能力强、收敛速度快的量子遗传算法（QGA）对模型参数进行参数寻优,把LSSVM参数选择问题转化为优化问题。该算法克服了遗传算法优化过程中陷入局部极值的问题,提高了优化性能。利用UCI数据库的数据进行分类验证,相比遗传优化的LSSVM和交叉验证的LSSVM,基于QGA优化的LSSVM模型提高了分类精度。最后,把该模型应用于风力发电机齿轮箱故障诊断中,取得了良好的效果。     

基于形状的CT图像断层间数据插值算法

CT脑血管三维重建图像都是源自二维断层扫描,由于二维断层扫描图像是对一定厚度的三维物体在二维平面上的断层投影,在目前的技术水平下,相邻两断层图像间的扫描距离要大于同一断层中解剖组织相邻像素间的图像显示距离.为保证三维重建图像的质量,在研究了CT脑血管二维图像特征后,提出了一种新的基于形状和基于灰度的图像断层间数据插值算法,并取得了较理想的实验效果.     

基于改进Harris-SIFT算子的快速图像配准算法

提出了一种基于改进Harris-SIFT算法的快速图像配准方法。首先,对传统Harris算法进行改进:一是构建高斯尺度空间,提取具有尺度不变性的角点特征;二是借鉴Forsnter算子思想对提取的角点进行精定位,以提高配准精度。然后,利用SIFT算子的特征描述方法对提取的特征点进行描述,通过随机kd树算法对两幅影像的特征点进行匹配。最后采用RANSAC算法对匹配点对进行提纯,并通过最小二乘法估计两幅影像间的空间几何变换参数,完成图像配准。实验结果表明:本文方法在基本保持配准精度的同时,能够大大减少标准SIFT所需的配准时间。     

基于降采样和改进Shi-Tomasi角点检测算法的PCB图像拼接

针对大型多重复单元PCB图像拼接耗时长、拼接错误率高等问题,提出了一种快速鲁棒的图像拼接方法.对采集到的高分辨率PCB图像进行降采样,基于人工选点精准获取含重叠区域的图像单元作为配准区域;引入抑制半径的方法对Shi-Tomasi角点检测算法进行改进,使提取出的区域特征点分布更加均匀;使用暴力匹配方式分别对区域特征点进行粗匹配并通过RANSAC算法剔除误匹配点对后获得配准系数矩阵;结合仿射变换公式推导计算出原图像的配准系数矩阵,根据配准系数矩阵对待拼接的图像进行融合,得到完整的PCB拼接图像.实验结果表明,所提出的PCB图像拼接方法,加快了PCB图像拼接的速度同时也提高了特征点匹配精度,在对图像降采样8倍下,改进的Shi-Tomasi算法较传统的Shi-Tomsi算法和Harris算法在匹配正确率上分别提高了7.8％和4.0％,验证了该方法的可行性.     

An MR-compatible stereoscopic in-room 3D display for MR-guided interventions

Background and methods

A commercial three-dimensional (3D) monitor was modified for use inside the scanner room to provide stereoscopic real-time visualization during magnetic resonance (MR)-guided interventions, and tested in a catheter-tracking phantom experiment at 1.5 T. Brightness, uniformity, radio frequency (RF) emissions and MR image interferences were measured.

Results and discussion

Due to modifications, the center luminance of the 3D monitor was reduced by 14 %, and the addition of a Faraday shield further reduced the remaining luminance by 31 %. RF emissions could be effectively shielded; only a minor signal-to-noise ratio (SNR) decrease of 4.6 % was observed during imaging. During the tracking experiment, the 3D orientation of the catheter and vessel structures in the phantom could be visualized stereoscopically.