基于轮廓的圆形仪表自动透视校正

针对利用机器视觉对圆形仪表进行读数识别的过程中,由于镜头平面与仪表平面不一定平行导致采集图像发生几何畸变的问题,提出一种基于轮廓的圆形仪表自动透视校正算法。该算法分为校正矩阵估计与图像校正两个部分,首先通过圆形仪表的轮廓来计算校正矩阵,再利用双线性插值校正畸变图像,该算法只需要获得圆形仪表轮廓的坐标集合就可以校正畸变图像。仿真实验结果表明,基于轮廓的校正算法与传统四点投影变换算法相比,校正精确度提高了2%～6%,同时实际仪表校正实验也表明轮廓校正算法有着更高的精确度以及稳定性,对圆形仪表的自动读数的预处理有实际应用价值。     

广角镜头畸变测量及校正方法研究

为了提高广角气溶胶探测仪的成像质量,提出了根据实验测量的拟合数据对图像进行畸变校正的方法。使用平行光管配合精密转台对探测仪光学系统的畸变量进行了实验测量,对畸变测量数据进行了多项式拟合,得到了光学系统在整个视场内的畸变量分布。实验结果表明,通过坐标变换与灰度校正,可以使探测仪采集图像的最大畸变量由10个像元偏移量(0.14 mm)减小到2个像元偏移量(0.028 mm)。采用基于实验测量畸变对所采集图像进行校正的方法,可以有针对性的对系统畸变进行校正,提高成像质量。     

针对鱼眼图像边缘校正失真的快速校正算法

针对现有鱼眼图像畸变校正算法校正后图像边缘区域存在失真的问题,在经度校正算法的基础上引入拉伸因子设计一种快速校正算法。首先利用快速扫描方法提取图像有效区域,得到畸变图像有效区域半径及圆心,其次在鱼眼图像畸变模型和径向畸变原理的基础上,确定畸变校正拉伸因子,然后分别在经度和纬度方向上增加拉伸因子进行校正,最终将校正结果与经度校正算法、经纬校正算法、双经度算法和重定心校正算法进行比较。结果表明,该算法能有效改善图片边缘的失真现象,且校正效率相比其它算法均有明显提高。综上所述,本文提出的算法能够快速有效地对鱼眼图像进行高质量的校正,为畸变校正质量提升相关研究提供借鉴。     

Correction of <Emphasis Type="Italic">B</Emphasis><Subscript>0</Subscript>-induced geometric distortion variations in prospective motion correction for 7T MRI

Objective

Prospective motion correction can effectively fix the imaging volume of interest. For large motion, this can lead to relative motion of coil sensitivities, distortions associated with imaging gradients and B ₀ field variations. This work accounts for the B ₀ field change due to subject movement, and proposes a method for correcting tissue magnetic susceptibility-related distortion in prospective motion correction.

Materials and methods

The B ₀ field shifts at the different head orientations were characterized. A volunteer performed large motion with prospective motion correction enabled. The acquired data were divided into multiple groups according to the object positions. The correction of B ₀-related distortion was applied to each group of data individually via augmented sensitivity encoding with additionally integrated gradient nonlinearity correction.

Results

The relative motion of the gradients, B ₀ field and coil sensitivities in prospective motion correction results in residual spatial distortion, blurring, and coil artifacts. These errors can be mitigated by the proposed method. Moreover, iterative conjugate gradient optimization with regularization provided superior results with smaller RMSE in comparison to standard conjugate gradient.

Conclusion

The combined correction of B ₀-related distortion and gradient nonlinearity leads to a reduction of residual motion artifacts in prospective motion correction data.

     

基于鱼眼相机畸变图像的大尺寸目标测量方法

为满足大尺寸目标高精度图像测量的需求,提出一种利用鱼眼畸变图像实现宽视野、大几何尺寸目标测量方法。根据非线性畸变的多项式逼近函数建立鱼眼相机成像模型,通过立体标定板来确定相机参数;将相机沿光轴方向移动采集一组图像,建立鱼眼相机的测量模型实现待测目标尺寸的测量。通过对棋盘格和某大型建筑物几何尺寸的测量实验表明,本文提出的直接利用畸变图像进行测量的方法优于传统的将畸变图像校正成线性图像测量的方法,具有更高的测量精度和稳定性。     

A method of RF inhomogeneity correction in MR imaging

A direct postprocessing method for correcting RF inhomogeneity in MR imaging is proposed. First, two images with different flip-angles of ψ and 2ψ are obtained. Next, the spatial distribution maps of the sensitivity of the surface coil and theB ₁ field intensity are produced by employing those images. Finally, the correction of the MR image is achieved, dividing the original image by distribution maps of the coil sensitivity and theB ₁ field intensity. The method was applied to image obtained by a gradient echo sequence and the corrected image is presented.     

基于强度-梯度映射与多方向中值滤波的红外弱小目标检测算法

为了能够在复杂环境下准确定位出弱小目标,依据目标的高斯形状特性,设计了强度-梯度映射耦合多方向中值滤波的 红外弱小目标检测算法。 首先,根据红外图像在 4 个不同方向的强度均值,对经典的中值滤波进行改进,以有效抑制复杂背景 中的噪声。 再基于弱小目标的中心像素,获取整个红外图像的强度信息。 将红外图像沿着半径方向分割为 4 个子块,并建立每 个子块的极坐标系统,以计算其对应的梯度值。 依据最大与最小梯度值的比率,得到整个红外图像的梯度信息。 再将强度与梯 度信息实施融合,得到背景抑制图像,以增强红外弱小目标。 最后,利用强度-梯度映射中的非零像素均值来计算阈值,对背景 抑制图像实施分割,准确定位弱小目标。 测试数据显示,与已有的红外弱小目标检测方案相比,所提算法具备更高的检测准确 性,可完整地识别出目标,呈现出更为理想的 ROC(receiver operating characteristic)曲线。     

注塑件机器视觉缺陷检测的几何矫正方法研究

针对多面体注塑零件在机器视觉缺陷检测中的零件图像几何形变问题,提出了基于几何光学原理的矫正算法。在拍摄定位误差不大于1 mm的条件下,所述方法矫正误差理论上<0.1 mm,可满足注塑零件机器视觉缺陷检测的需要。首先对采集的图像进行预处理获取图像边缘;接着将轮廓线交点确定为零件顶点;根据顶点位置分割零件的不同表面并将其映射在二维平面;然后根据几何光学计算图像中每一个像素点的偏移量;最后使用基于几何光学的方法对图像中的像素点进行逐点矫正。利用一组六面体零件模拟实际工况,在不同的拍摄定位误差状态下进行实验,使用Matlab对矫正算法进行验证。实验结果表明,所述方法误差在0.1 mm以内,与理论分析相吻合,满足注塑零件在机器视觉缺陷检测中零件图像几何矫正精度的需要。     

Highly accelerated PSF-mapping for EPI distortion correction with improved fidelity

Objective

This study presents an improved point-spread-function (PSF) mapping-based distortion correction method and accelerated PSF acquisition for distortion correction in EPI without loss of quality or reliability compared to full encoding.

Materials and methods

To correct geometric distortions accurately, the PSF in the EPI phase-encoding coordinates (EPI-PSF) was measured and used as a kernel for distortion correction. FOV reduction was applied in the PSF mapping dimension for highly accelerated PSF acquisition. A novel approach for fold-over artifact correction in this reduced dimension is introduced. Conventional gradient-echo EPI and corresponding full PSF reference data were acquired in phantoms and in human brain at 7 T. The distortion corrected EPI data with the proposed acceleration were compared to result with full encoding. Previously published interpolation methods based on shift maps, non-uniform Fourier transformation and a b-spline interpolation were compared with the proposed method.

Results

The results demonstrate that the proposed method corrects geometric distortions in EPI with high accuracy and quality despite the high acceleration. In contrast to partial parallel imaging acceleration, no noise enhancement is introduced.

Conclusion

The proposed EPI-PSF-based distortion correction improves correction of EPI and accelerates PSF reference data acquisition and computation.     

3D gradient system for two B 0 field directions in earth’s field MRI

Object

A new gradient system for earth’s field magnetic resonance imaging (EFMRI) is presented that can be rotated relatively to the earth’s field direction while maintaining the ability to encode images. Orthogonal components of the gradient field are exploited to reduce the number of gradient coils.

Materials and methods

Two favorable orientations of the gradient system relative to the earth’s magnetic field (parallel and perpendicular) are discussed. We introduce the theory for the magnetic fields of the new gradient system and illustrate the design of the coil geometries which were worked out with the help of simulations and a numerical optimization algorithm. Field mapping measurements and imaging experiments in the two different orientations of the gradient system were carried out.

Results

Orthogonal components of the gradient field take over the role of the additionally needed gradient fields when the gradient system is rotated relative to the earth’s magnetic field. The results from the field mapping and imaging experiments verify the presented theory and show the functionality of the new gradient system.

Conclusion

The presented system demonstrates that gradient coils can be used for image encoding in multiple directions. This fact can be exploited to realize an EFMRI setup for parallel and perpendicular prepolarization with a single set of gradient coils.     

基于统计特征分类耦合自适应 Gamma校正的图像增强算法

为了避免图像在亮度增强时导致其颜色失真,且在局部易出现过增强等问题,设计了统计特征分类耦合自适应Gamma校正(adaptive Gamma correction,AGC)的图像增强算法,以更好提高图像细节与视觉效果。首先,将输入图像转换为HSV空间,使颜色与亮度分离,使其在增强亮度通道时不改变像素的原始颜色,有效降低颜色失真。然后,考虑不同图像的性质,利用统计信息将图像分类为高、低两种对比度,每种对比度又分为亮、暗两类。其次,基于传统的Gamma校正方法,通过对于不同类型的图像进行动态参数设置,形成一种AGC机制,从而为不同类型图像的构建了不同的增强函数,以完成不同类别图像的增强处理。实验数据表明,与当前流行的增强算法相比,所提算法具备更高的增强效果,呈现出更为自然的亮度与对比度,且保持了更多的颜色信息。     

激光标刻系统几何失真的校正方法

激光标刻系统的复合几何畸变并非其固有缺陷,畸变产生的原因在于简化的、近似的系统控制策略.依据扫描振镜的扫描几何公式,采用最速下降法解算了扫描振镜偏转角度的数值解,实现了目标标刻坐标到扫描振镜偏转角度的精确映射.在扫描振镜偏转角度数值解的基础上,利用扫描振镜复合几何畸变的对称特性性,通过在第1象限建立适当规模的误差校正表并采用插值算法进行失真校正.仿真实验结果表明,该方法能够在4象限实现满足精度要求的失真校正.通过将目标标刻坐标转换为归一化坐标,使得该校正方法对于不同焦距的扫描振镜具有广泛的适用性.     

激光标刻系统几何失真的校正方法

激光标刻系统的复合几何畸变并非其固有缺陷,畸变产生的原因在于简化的、近似的系统控制策略。依据扫描振镜的扫描几何公式,采用最速下降法解算了扫描振镜偏转角度的数值解,实现了目标标刻坐标到扫描振镜偏转角度的精确映射。在扫描振镜偏转角度数值解的基础上,利用扫描振镜复合几何畸变的对称特性性,通过在第I象限建立适当规模的误差校正表并采用插值算法进行失真校正。仿真实验结果表明,该方法能够在四象限实现满足精度要求的失真校正。通过将目标标刻坐标转换为归一化坐标,使得该校正方法对于不同焦距的扫描振镜具有广泛的适用性。     

自适应水下彩色图像增强算法

在水下环境中,水体介质对光照的吸收和散射是导致水下成像失真的两大主要成因。水体对光线的吸收导致能量衰退大大降低了清晰度,另水体介质对各波段的散射和吸收的无规律性导致水下成像颜色失真。为了提高对水下成像的认知,本文提出了一种自适应水下图像增强算法。该方法首先对原始图像进行降噪和对比度提升得到两幅输入图;其次,以原始图像的显著图为引导,结合照度和色度图,作为图像融合的权重图;最后,根据权重图对两幅输入图像进行自适应加权求和,从而实现图像增强的目的。实验结果证明本文算法具有良好的实时性和鲁棒性,有效地增强了水下降质图像。     

彩色宽幅面阵航空相机的非均匀性校正方法

分析某型Bayer阵列彩色宽幅面阵航空相机的成像过程,结合Bayer阵列成像原理,提出相机成像均匀性的各个影响因素。采用实验室内辐射标定的方法获得相机暗电流噪声及像元响应度的测量数据,响应度测量结果表明,在工作谱段范围内,当输出值未达到饱和时,相机各色彩通道像元的响应度曲线接近线性变化,但像面边缘区域的像元响应度较中心区域明显降低。以响应度最高的红色通道为基准,获得等能量输入时各色彩通道的归一化色彩平衡系数,结合各像元响应特性,提出基于辐射亮度反演和等能量色彩平衡系数的非均匀性校正方法,并对相机航空成像图像进行了非均匀性校正,均值、方差及梯度等统计结果表明,校正后图像各个区域各个色彩通道的灰度均值差异明显减小,尤其是红色通道,其左下区域标准方差比由0.201 5降低到0.076 1,说明图像整体亮度更均衡,色调更均匀,校正后图像各个色彩通道的平均梯度较原始图像有所提高,说明校正后图像细节信息更加丰富。     

基于改进亚像素算法的陶瓷天线PIN针在线精密检测

陶瓷天线PIN针作为天线中重要的零部件之一,其尺寸偏差将直接关系到天线的产品质量。为了实现PIN针的在线精密检测,设计研发了一种PIN针在线检测装置,并提出了一种基于改进亚像素算法的PIN针尺寸检测方法。首先开始采集图像并进行像素当量标定,对图像进行畸变校正、获取ROI区域、图像预处理,然后利用基于改进Sobel算子及高斯峰值位置估算的亚像素边缘检测算法提取边缘点,利用最小二乘法将边缘点拟合成一对平行直线,并计算出线间像素宽度,根据像素当量换算得到被测PIN针在ROI区域处的直径尺寸。实验结果表明,该方法的测量平均相对误差小于0.25%,在保证±0.005 mm检测精度的同时,其平均耗时相对于传统基于高斯拟合的亚像素检测算法缩短了64.32%。     

基于质量度量与颜色校正的多曝光图像融合算法

针对当前多曝光图像融合过程的图像质量属性选择不当,易导致颜色失真与丢失细节等问题,设计了一种质量度量耦合颜色校正的多曝光图像融合算法。首先,选择3种最显著的图像质量属性(对比度、饱和度和亮度)作为度量方式;然后,利用线性组合将得到的3种质量属性进行成加权组合,并采用幂函数来控制每个属性的影响,将较低的权重值赋予曝光不足和曝光过度的像素,消除具有视觉效果不佳的像素,从而有效保留正常的曝光像素及其明亮的颜色与细节;随后,将不同曝光图像加权组合的特征进行Laplacian金字塔分解,经过归一化权重映射后,对其进行不同系数的多分尺度融合,完成多曝光图像融合。为了避免颜色失真与细节丢失,采用基于局部饱和度的颜色校正方法来改善图像质量。实验数据表明,与当前多曝光图像融合方案相比,所提算法具有更高的融合视觉质量,可以更好地保持图像细节和校正曝光融合图像的颜色。     

基于全景相机的3D坐标测量方法

全景相机通过一次成像获取其周围3600场景,是一种大视场快速图像采集装置。研究了应用双鱼眼全景相机获取三维空间点的方法,研究了单鱼眼镜头成像模型,双鱼眼全景相机成像模型及图像变换原理。研究了鱼眼镜头成像畸变等校模型参数的标定方法,实现了对鱼眼畸变图像的矫正。提出了由全景图像数据解算任意方向等效鱼眼镜头成像数据的数学模型。最后应用平行光轴双目立体视觉技术,获得了被测场景的3D坐标数据。研究结果表明,应用这种方法在4 m×3 m×3 m的测量范围内精度可达13 mm。     

Design and fabriacation of a three-axis multilayer gradient coil for magnetic resonance microscopy of mice

There is great interest in the non-destructive capabilities of magnetic resonance microscopy for studying murine models of both disease and normal function; however, these studies place extreme demands on the MR hardware, most notably the gradient field system. We designed, using constrained current minimum inductance methods. and fabricated a complete, unshielded three-axis gradient coil set that utilizes interleaved, multilayer axes to achieve maximum gradient strengths of over 2000 mT m⁻¹ in rise times of less than 50 μs with an inner coil diameter of 5 cm. The coil was wire-wound using a rectangular wire that minimizes the deposited power for a given gradient efficiency. Water cooling was also incorporated into the coil to assist in thermal management. The duty cycle for the most extreme cases of single shot echo planar imaging (EPI) is limited by the thermal response and expressions for maximum rates of image collection are given for burst and continuous modes of operation. The final coil is capable of the collection of single shot EPI images with 6 mm field of view and 94 μm isotropic voxels at imaging rates exceeding 50 s⁻¹.     

基于毕奥-萨伐尔定律的永磁球形电动机定子线圈磁场分布研究

永磁球形电动机定子绕组采用集中式绕制方式,分层均匀镶嵌在定子球壳内壁。为得到永磁球形电动机单个定子线圈在三维空间中的磁场分布,建立球形电动机全局坐标系与定子线圈局部坐标系之间的映射关系,然后运用毕奥-萨伐尔定律对定子线圈产生的磁场进行求解分析。在对磁场解析表达式求解过程中借鉴“割圆法”思想,得到了磁场的数值解表达式,给出了单个定子线圈在三维空间中的磁场分布,并通过有限元方法验证了所提方法的有效性。