一种基于多项式插值改进的亚像素细分算法

介绍了一种基于多项式插值法的改进的亚像素细分算法及相应计算模板和公式,并对算法的误差进行了分析.本算法应用在经典Sobel算子基础上构造出的方向模板,对灰度图像进行处理,得到梯度图像,然后在梯度图像上沿目标边缘的梯度方向进行多项式插值法亚像素细分计算,对目标边缘进行亚像素精确定位.实例说明本算法是可行的.     

尺寸测量中的边缘检测算法研究

主要介绍了一种改进的边缘检测算法及相应计算模板和公式,并对算法的误差进行了分析。因为传统的边缘检测算法在理论上较成熟,但在实际应用时可操作性较差。所以针对零件图像测量的实际应用情况,使用基于Sobel算子的改进的方向算子,对灰度图像进行处理得到梯度图像。综合应用其他算法,实现了对目标边缘的准确检测。然后在梯度图像上沿目标边缘的梯度方向进行多项式插值法亚像素细分计算,对目标边缘进行亚像素精确定位。在文章的最后,用实例说明了本算法的可行性。     

自动影像测量系统关键算法

为了实现对工件的自动影像测量,建立了自动影像测量系统。对该系统所采用的图元识别、图元测量、基于自动聚焦原理的高度测量等算法进行研究。根据圆形和矩形图元的面积和真圆度等特征参数介绍了图元识别算法。以直线和圆形图元为例分析了典型图元的测量算法,即在提取图元边缘点的基础上进行图元拟合。在分析比较能量谱等聚焦评价方法的性能的基础上,说明了采用自动聚焦原理进行高度测量的算法。最后,介绍了系统比例尺的设定方法以及光栅尺读数的误差补偿方法。实验结果表明:比例尺的标定精度为0.5 μm;图元的测量精度在微米级;高度测量精度为0.035 mm,基本满足自动影像测量的稳定可靠、精度高、抗干扰能力强等要求。     

Aligning content rendering resolution and feature size with display capability in near‐eye display systems

In order to create a satisfying experience with near‐eye displays, the content must be adapted to be legible on the display used. New displays are using subpixel arrangements that can limit the minimum resolvable feature size to something greater than with the conventional RGB stripe arrangement. We conducted an experiment to measure the minimum and preferred size of text in two virtual reality (VR) displays systems and find that the text size is display limited. We then measure several displays with different pixel arrangements to determine whether the subpixel arrangement could impact legibility. We propose several Fourier metrics that can be computed from the measured data to categorize the capability of the display and describe a framework for selecting the appropriate content from a set of discrete tiers.     

基于变结构元与参数映射的石英摆片参数测量

为了满足诸如民用领域和国防技术领域中对挠性器件石英摆片参数高精度测量的要求,提出一种利用视觉图像对石英摆片参数进行高精度测量的非接触方法.该方法通过视觉测量系统对石英摆片图像进行摄取,利用变结构元广义形态学边缘检测算法进行初始边缘定位,充分提取图像边缘细节信息的同时抑制图像噪声的影响.在亚像素图像处理中,提出了对圆形曲线进行亚像素边缘定位的边缘检测算法,建立了曲线边缘点与边缘参数之间的映射关系.利用检测出的曲线边缘点数据,通过定义一个新的误差函数并最小化,可以计算出石英摆片参数.实验结果表明,该方法稳定性好且实时性强,定位不确定度优于0.03像素,可实现石英摆片参数的精密测量.     

基于Zernike矩亚像素边缘检测的快速算法

为了克服传统的Zernike法在边缘检测过程中,由于人工手动选取阈值而带来的低效率、高误判等不足,将原算法与Otsu法相结合,提出了一种边缘检测的快速算法。利用传统的Zernike法计算出图像的阶跃灰度矩阵,再将该矩阵作为计算对象,用Otsu法直接得到最优的阶跃灰度阈值进行边缘判别,并考虑了由于边缘模型带来的误差,在保证检测效果的同时缩短了检测时间。实验结果表明,改进的算法能够更有效地完成边缘检测,补偿后的亚像素定位更准确。     

Sensitivity limits of strain mapping procedures using high-resolution electron microscopy

Strain detection limits achieved by several approaches of strain mapping differing in the procedure of determining positions of intensity maxima have been compared. Algorithms which simultaneously take into account the two-dimensional intensity distribution around maximum positions yield similar detection limits and are superior to those performing successive one-dimensional cuts. Investigations of the effect of image recording imply that the use of photographic plates is superior to direct image recording using a slow-scan CCD camera because of larger sampling rates and a larger field of view. It is finally shown that the choice of specimen preparation technique substantially affects the strain detection limit with cleaved samples yielding results which are a factor of about three better than ion-milled samples.     

The particle image tracking technique: An accurate optical method for measuring individual kinematics of rigid particles

We present a new approach to assess the two-dimensional motion of rigid bodies in granular materials. Although it was adapted from digital image correlation technique, the heart of the presented technique relies on specific treatments related to the discrete nature of grain-displacement fields. The code called Tracker has been developed to process the digital images and measure the in-plane displacement and rotation of each individual grain from one image to another. A remarkable feature is the use of a specific strategy that allows tracking all particles, without losing any of them (which is a typical problem when tracking assemblies of discrete particles over many images). This is achieved by a two-step procedure, where, in case of problematic tracking of a grain, the size of the search zone is increased in an adaptive manner, that is, taking into account the results of tracking in the neighbourhood of the particle. The accuracy of the measured displacements and rotations was tested on both perfect synthetic images and digital photographs of a sheared assembly of grains. An automatic procedure that corrects the lens distortion further improves the quality of the measurements. The accurate assessment of the grain kinematics opens very interesting perspectives, especially in the study of displacement fluctuations in granular media.     

亚像素位移插值计算方法的比较分析

介绍了亚像素数字散斑测量中常用的高斯曲面插值、抛物面插值及梯度插值原理，推导了相关的公式，提出了一种利用计算机图像处理软件考查数字散斑测量软件精度的方法，并利用这种方法对上述三种插值方法进行了比较分析．结果表明，抛物面插值法计算速度较快，精度较高，并且误差波动较小．     

基于多尺度密集连接和U-Net改进的动态场景去模糊算法

针对现有去运动模糊网络在图像恢复过程中出现的纹理细节丢失、无法抑制噪声、产生振铃伪影等问题,提出一种基于多尺度密集连接和U-Net改进的动态场景去模糊算法。首先,借助U-Net网络中空洞卷积下采样有效扩大感受野,在不增加参数量的情况下避免图片产生不可逆损伤,并利用亚像素卷积在上采样过程中以小的卷积核获得清晰的图像细节,降低运算复杂度;其次,设计多尺度密集特征提取模块(multi-scale dense feature extraction, MDFE),通过密集连接的卷积层加强深层次特征提取和复用,运用空间金字塔池化(spatial pyramid pooling, SPP)分支引导多尺度特征的传递和融合,促进图像细节纹理的有效保留;最后,采用ConvLSTM双向连通结构(bidirectional convolution LSTM unit, BCLU)以非线性方式从编码路径补偿简单级联流失的上下文特征,推动深度特征跨阶段相互作用,弱化边缘伪影和噪声干扰。与现有先进方法对比,验证了本文所提算法在性能上的优势。