基于B样条的真实感头发建模

引入B样条对头发及发型进行真实感建模与仿真.由悬臂梁模型生成初始数据点,经B样条曲线反算生成自然弯曲的关键头发;由周期性闭合B样条曲线反算生成头发簇的截面曲线;利用B样条的局部性质对关键头发和截面曲线进行形状调整;在关键头发和截面曲线的基础上运用蒙面算法生成头发簇.最后在头部模型上按区域生成头发簇并进行光照渲染完成整体发型设计与仿真.结果表明,该方法对头发的控制比较简单,可以获得各种发型.     

子空间约束的最小二乘常模算法

常模算法(CMA)是一种性能优良的盲多用户检测算法.最小二乘常模算法(LSCMA)因其全局收敛性和稳定性而备受关注,但是在信噪比低时性能不理想.本文将最小二乘常模算法,紧缩近似投影子空间(PASTd)算法和奇异值分解(SVD)相结合,提出了一种子空间约束的最小二乘常模算法,称为SUB_LSCMA,其复杂度比已有的基于直接对接收信号自相关矩阵做特征值分解(ED)的LSCMA_SUB[6]复杂度低.仿真结果表明这种算法的收敛速度、跟踪性能和误码性能和LSC-MA_SUB基本相同.     

红外弱小目标背景抑制中的二维空域廓线法

为了实现天基红外遥感图像中的弱小目标检测,抑制复杂的结构背景,提出了二维空域廓线法。算法以一维时域廓线目标检测理论为基础,将其扩展到二维空域,应用于单帧目标检测。采用二维空域廓线可以获得背景估计,用原始图像减去背景图像可获得包含目标的前景图像。将本文算法与现有的单帧目标检测算法相比,可获得更大的ROC曲线面积,表明提出算法优于同类算法。     

融合多元经验模态分解与Hilbert空间填充曲线的sEMG处理方法

基于卷积神经网络(convolutional neural network, CNN)的表面肌电信号(surface electromygraphy, sEMG)手势识别算法通常将一维sEMG转换成二维肌电图作为CNN的输入。针对sEMG瞬时样本量偏少、以及一维sEMG转换成二维肌电图时带来的局部时序特征丢失等问题,提出了将多元经验模态分解(multivariate empirical mode decomposition, MEMD)算法与Hilbert空间填充曲线相结合的方法,以提升手势识别算法的准确率。采用开源数据集NinaPro-DB1作为实验数据集;通过MEMD算法对sEMG进行分解;将分解后的本征模态函数(intrinsic mode functions, IMFs)作为Hilbert曲线的填充域(Hilb-IMFs)映射成二维肌电图;选择DenseNet作为手势识别的基本网络。实验结果表明,提出的方法相对于传统信号升维方法在手势识别准确率上约有4%的性能提升,验证了该方法的有效性。     

基于子空间方法的最小二乘常模算法的研究

本文提出了两种基于子空间方法的常模算法,称为SUB_LSCMA和LSCMA_PASTd。SUB_LSCMA先采用奇异值分解(SVD)获得紧缩近似投影子空间(PASTd)算法的初值,用PASTd算法来计算信号子空间,并对该信号子空间作施密特正交化,将最小二乘常模算法(LSCMA)的权系数投影到正交的信号子空间上,目的是减轻噪声子空间干扰的影响,但复杂度比已有的基于直接对接收信号自相关矩阵做特征值分解(ED)的LSCM_SUB算法[6]复杂度低。LSCMA_PASTd在SUB_LSCMA的基础上作了进一步改进,采用改进的PASTd算法来计算信号子空间,该信号子空间具有正交性,并且对初值的选取不敏感,能运用于实际的多径衰落信道中。仿真结果表明这两种算法的收敛速度、跟踪性能和误码性能和LSCM_SUB算法基本相同,但是复杂度比LSCM_SUB算法低。     

采用格林函数分解的太赫兹逆合成孔径雷达近场成像算法

针对太赫兹逆合成孔径雷达(ISAR)近场成像的问题,本文提出了采用格林函数分解的ISAR成像算法。文中结合太赫兹雷达超外差接收模式,构建和分析了近场条件下的目标回波模型。通过对回波模型中格林函数的分解,将表征球面波前的格林函数分解为二维平面波的叠加;并在频域完成对回波信号的滤波补偿和二维逆变换后,获得了太赫兹ISAR近场目标的图像。文中通过与二维FFT算法和本文算法进行的数值仿真结果对比,验证了该算法的可行性。最后利用太赫兹雷达转台实测数据对本文算法进行了验证和对比分析,结果表明该算法具有良好的成像性能。      

基于一类支持向量机的冠脉病变检测方法

针对冠脉病变检测算法普遍存在的异常截面识别率低、无法排除特殊结构影响等问题,提出了一种基于一类支持向量机(OCSVM)的冠脉病变检测方法,并使用冠脉面重采样和基于最大互信息的特征选择方法提高了算法识别正确率。该方法首先基于梯度通量对冠脉源截面进行三次样条插值重采样,然后构造出截面的多尺度特征,接着使用最大互信息结合冗余度去除进行特征选择,最后使用特征数据训练OCSVM完成冠脉病变检测。实验结果显示,在1128个冠脉截面数据的测试结果中,本算法在完全识别异常截面的情况下对健康截面的识别正确率达到了53.5%,远高于同类型的仅从正面和未标记数据学习的支持向量机(SVM)算法所对应的19.6%;而冠脉截面重采样也使得30个特征数下算法对健康截面的识别正确率由21.7%提高到了53.2%。     

基于ISAR的曲线SAR自聚焦算法

在曲线合成孔径雷达(synthetic aperture radar,SAR)系统中,由于采用曲线孔径使得孔径误差补偿问题十分复杂.本文基于逆合成孔径雷达(inverse SAR,ISAR)相位补偿思想,用参考点所在距离单元相位补偿孔径误差,并由补偿数据中提取散射点的三维特征.仿真实验表明,新算法由无噪数据中获得的散射点分布与散射点真实分布只是存在着空间位置上的偏移,但目标结构特征未变.因此新算法是一种适用于曲线SAR的有效自聚焦算法.     

神经网络在自适应噪声抵消中的应用研究

本文将基于正交最小二乘的RBF神经网络算法引入自适应噪声对消中,提出一种基于最小二乘算法和径向基网络的自适应噪声抵消(adaptive filter based on least square algorithm and radial basis network,简称OLSRBFAF)算法。RBF网络因其具有良好的推广能力,简单的结构和快速的训练过程等诸多优点已被成功应用于很多领域。RBF神经网络中关键因素是基函数中心的选取,中心选取不当构造出来的RBF网络的性能一般不能令人满意。利用正交最小二乘(orthogonal least squares,简称OLS)算法选取RBF网络中心,解决了径向基函数网络构造这一关键问题。并由于OLS算法中采用了最小二乘(least-square,简称LS)准则,其对时变信道具有快速跟踪的能力。利用MATLAB仿真结果分析可知,通过将两种算法结合引入自适应噪声抵消系统,使该系统具有误差更小,消除噪声能力更强的优点。     

用可变形模板进行基于内容的图像分割算法

 本文提出一种采用可变形模板匹配技术进行基于内容的图像分割算法.通过预先计算出可变形模板沿着变形的正交曲线,并对模板曲线及正交曲线进行离散抽样,建立一基于正交曲线的二维(2-D)可变形模板,针对图像分割问题定义控制可变形模板进行变形的内、外部能量函数,本文采用遗传算法搜索能量函数最小的全局最优解.该新算法比传统的可变形模板匹配方法降低了搜索空间的维数,减少了算法对模板初始位置的敏感.对实际图像及模拟低信噪比图像处理的结果表明,新算法具有良好的分割精度及稳定性.     

基于波前恢复的温度场真实三维重建

提出了采用傅里叶分析和改进的多重网格法恢复多方向干涉波前的技术,得到待测场在不同方向上的二维投影数据,从而实现了任意截面的场重建,即真正的三维重建。并实验利用了一种旋转F-P干涉仪,实时捕获非对称温度场的多方向干涉图,并给出由这些二维投影重建三维分布的结果,证明此采集、处理及重建方法是行之有效的。     

Surface interpolation from sparse cross sections using region correspondence

The ability to estimate a surface from a set of cross sections allows calculation of the enclosed volume and the display of the surface in three-dimensions. This process has increasingly been used to derive useful information from medical data. However, extracting the cross sections (segmenting) can be very difficult, and automatic segmentation methods are not sufficiently robust to handle all situations. Hence, it is an advantage if the surface reconstruction algorithm can work effectively on a small number of cross sections. In addition, cross sections of medical data are often quite complex. Shape-based interpolation is a simple and elegant solution to this problem, although it has known limitations when handling complex shapes. In this paper, the shape-based interpolation paradigm is extended to interpolate a surface through sparse, complex cross sections, providing a significant improvement over our previously published maximal disc-guided interpolation. The performance of this algorithm is demonstrated on various types of medical data (X-ray computed tomography, magnetic resonance imaging and three-dimensional ultrasound). Although the correspondence problem in general remains unsolved, it is demonstrated that correct surfaces can be estimated from a limited amount of real data, through the use of region rather than object correspondence.     

Reconstructing the cross sections of coronary arteries from biplane angiograms

An algorithm that reconstructs the cross sections of the lumens of coronary arteries from two mutually orthogonal X-ray projections is described. The algorithm accommodates the possibility of elliptical, crescent, or star shapes. It represents each biplane projection of a transverse slice of the arterial lumen as a binary-valued image. The single-coordinate moments of these two projection images are equal to those of the slice. Since the cross-coordinate moments of the slice are not available from the projections, an algorithm to estimate these moments based on assumptions of smoothness and connectivity is developed. Once all the missing moments are estimated, the image of the slice can be estimated by inverting these moments, using the uniqueness theorem governing the relation between an image and its moments. Preliminary tests of the algorithm on synthetic data, on hardware phantoms and on a segment of a barium-enhanced in vitro coronary artery are reported.     

Nevoscopy: three-dimensional computed tomography of nevi and melanomas in situ by transillumination

The thickness of a malignant nevus has been found to be an important prognostic factor for patients with melanoma. We have designed a new method of imaging nevi that permits their thickness to be measured in situ. Using fiber optics directed into the surrounding skin, we transilluminate the nevus. Three images are picked up by a digitizing TV camera: the vertical image (90 degrees ), a glancing image (180 degrees ), and one at 45 degrees , obtained by using two front-silvered mirrors held next to the nevus in a "nevoscope." The digitized images are used in a computed tomography algorithm to calculate approximate vertical cross sections of the nevus. The algorithm is one we recently developed to permit reconstruction from a very few projections. Our method is completely noninvasive. It may be used to check all the nevi on a patient. Without excisions, we could establish a baseline three-dimensional shape for each nevus, follow any changes in time, and obtain an early warning of increase in thickness and possible malignancy.     

Respiratory compensation in projection imaging using a magnification and displacement model

Respiratory motion during the collection of computed tomography (CT) projections generates structured artifacts and a loss of resolution that can render the scans unusable. This motion is problematic in scans of those patients who cannot suspend respiration, such as the very young or intubated patients. Here, the authors present an algorithm that can be used to reduce motion artifacts in CT scans caused by respiration. An approximate model for the effect of respiration is that the object cross section under interrogation experiences time-varying magnification and displacement along two axes. Using this model an exact filtered backprojection algorithm is derived for the case of parallel projections. The result is extended to generate an approximate reconstruction formula for fan-beam projections. Computer simulations and scans of phantoms on a commercial CT scanner validate the new reconstruction algorithms for parallel and fan-beam projections. Significant reduction in respiratory artifacts is demonstrated clinically when the motion model is satisfied. The method can be applied to projection data used in CT, single photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance imaging (MRI).     

A quantitative analysis of 3-D coronary modeling from two or more projection images

A method is introduced to examine the geometrical accuracy of the three-dimensional (3-D) representation of coronary arteries from multiple (two and more) calibrated two-dimensional (2-D) angiographic projections. When involving more then two projections, (multiprojection modeling) a novel procedure is presented that consists of fully automated centerline and width determination in all available projections based on the information provided by the semi-automated centerline detection in two initial calibrated projections. The accuracy of the 3-D coronary modeling approach is determined by a quantitative examination of the 3-D centerline point position and the 3-D cross sectional area of the reconstructed objects. The measurements are based on the analysis of calibrated phantom and calibrated coronary 2-D projection data. From this analysis a confidence region (alpha degrees approximately equal to [35 degrees - 145 degrees]) for the angular distance of two initial projection images is determined for which the modeling procedure is sufficiently accurate for the applied system. Within this angular border range the centerline position error is less then 0.8 mm, in terms of the Euclidean distance to a predefined ground truth. When involving more projections using our new procedure, experiments show that when the initial pair of projection images has an angular distance in the range alpha degrees approximately equal to [35 degrees - 145 degrees], the centerlines in all other projections (gamma = 0 degrees - 180 degrees) were indicated very precisely without any additional centering procedure. When involving additional projection images in the modeling procedure a more realistic shape of the structure can be provided. In case of the concave segment, however, the involvement of multiple projections does not necessarily provide a more realistic shape of the reconstructed structure.     

Characterization of optimum polarization for multiple targetdiscrimination using genetic algorithms

A stochastic optimization algorithm is used to characterize the polarization states of a nonpolarimetric radar transmitter and receiver antennas for optimal target classification. Specifically, the optimized solution is sought when a multitude of targets are to be categorized. It is shown that the objective function of the optimization problem is highly nonlinear and discontinuous, hence, classical optimization algorithms fail to provide satisfactory results. The stochastic optimization algorithm used is based on a genetic algorithm (GA) which operates on a discretized form of the parameter space and searches globally for the optimum point. In this process, it is assumed that the polarimetric responses of the targets are known a priori. The optimization algorithm is applied to two sets of data: (1) a synthetic backscatter data for four point targets with similar radar cross sections (RCSs) and (2) a set of polarimetric backscatter measurements of asphalt surfaces under different physical conditions at 94 GHz. The purpose of the latter study is to come up with the optimal design for polarization states of an affordable millimeter-wave radar sensor that can assess traction of road surfaces     

BOTDA传感系统的布里渊频移提取方法研究进展

随着布里渊光时域分析(BOTDA)传感技术在许多大型基础工程设施安全监测中的广泛应用,对测量精度和实时性的要求日益提高.采用传统的最小二乘曲线拟合方式对布里渊散射谱进行布里渊频移提取,其测量结果的精度依赖于参数初始值的选取和噪声的影响,并且拟合算法的参数迭代求解过程增加了数据处理的时间,降低了工程实时性.文章综述了多种非线性参数优化估计的曲线拟合算法和基于神经网络的布里渊散射谱特征提取的混合优化算法,介绍了无需经过曲线拟合的互相关法(XCM)、深度学习法(DL)和亚像素级精度的重心提取算法(CDA),这些算法能适应更大的扫频步长,实时性更好.     

光学元件面形误差的光滑延展算法

在离子束抛光工艺中,为了提高驻留时间求解算法在工件边缘处的求解精度,需要对原始面形误差数据进行有界光滑延展。推导并提出了一种基于高斯曲线的曲面延展算法,该算法利用了高斯函数的有界性、光滑性和参数连续性。将该曲面延展算法应用于带有高频噪声的面形误差工件的驻留时间求解过程中,驻留时间算法在光学元件边缘处的残余误差得到了抑制,使得驻留时间算法在整个通光口径内的收敛率达到了97%（RMS）。这表明基于高斯曲线的曲面光滑延展算法能实现光学元件面形误差的光滑延展,并具有良好的抗噪声干扰能力,改善了驻留时间算法的求解精度。     

A probabilistic framework based on hidden markov model for fiducial identification in image-guided radiation treatments

Fiducial tracking is a common target tracking method widely used in image-guided procedures such as radiotherapy and radiosurgery. In this paper, we present a multifiducial identification method that incorporates context information in the process. We first convert the problem into a state sequence problem by establishing a probabilistic framework based on a hidden Markov model (HMM), where prior probability represents an individual candidate's resemblance to a fiducial; transition probability quantifies the similarity of a candidate set to the fiducials' geometrical configuration; and the Viterbi algorithm provides an efficient solution. We then discuss the problem of identifying fiducials using stereo projections, and propose a special, higher order HMM, which consists of two parallel HMMs, connected by an association measure that captures the inherent correlation between the two projections. A novel algorithm, the concurrent viterbi with association (CVA) algorithm, is introduced to efficiently identify fiducials in the two projections simultaneously. This probabilistic framework is highly flexible and provides a buffer to accommodate deformations. A simple implementation of the CVA algorithm is presented to evaluate the efficacy of the framework. Experiments were carried out using clinical images acquired during patient treatments, and several examples are presented to illustrate a variety of clinical situations. In the experiments, the algorithm demonstrated a large tracking range, computational efficiency, ease of use, and robustness that meet the requirements for clinical use.