Remote-controlled flexible pose measurement system and method for a moving target in wind tunnel

The measurement of position and attitude parameters for the isolated target from a high-speed aircraft is a great challenge in the field of wind tunnel simulation technology. This paper proposes a remote-controlled flexible pose measurement system in wind tunnel conditions for the separation of a target from an aircraft. The position and attitude parameters of a moving object are obtained by utilizing a single camera with a focal length and camera orientation that can be changed based on different measurement conditions. Using this proposed system and method, both the flexibility and efficiency of the pose measurement system can be enhanced in wind tunnel conditions to meet the measurement requirements of different objects and experiments, which is also useful for the development of an intelligent position and attitude measurement system. The position and the focal length of the camera also can be controlled remotely during measurements to enlarge both the vertical and horizontal measurement range of this system. Experiments are conducted in the laboratory to measure the position and attitude of moving objects with high flexibility and efficiency, and the measurement precision of the measurement system is also verified through experiments.     

基于Mecanum轮配送车的视觉与磁带双重导引技术研究

针对卫星总装过程中提出的物料配送需求,设计并实现了以Mecanum轮全方位移动平台为基础,可在狭窄配送通道内快速、灵活移动的智能物料配送车。结合总装现场实际环境条件,配送车采用了视觉导引与磁带导引相结合的智能导引策略,有效消除了光照强度变化、地面裂纹、电磁信号干扰等环境因素对车道线采集的影响,最终实现总装物料从库房至多工位的自动配送。     

Visual inspection of aircraft skin: Automated pixel-level defect detection by instance segmentation

Skin defect inspection is one of the most significant tasks in the conventional process of aircraft inspection. This paper proposes a vision-based method of pixel-level defect detection, which is based on the Mask Scoring R-CNN. First, an attention mechanism and a feature fusion module are introduced, to improve feature representation. Second, a new classifier head—consisting of four convolutional layers and a fully connected layer—is proposed, to reduce the influence of information around the area of the defect. Third, to evaluate the proposed method, a dataset of aircraft skin defects was constructed, containing 276 images with a resolution of 960 × 720 pixels. Experimental results show that the proposed classifier head improves the detection and segmentation accuracy, for aircraft skin defect inspection, more effectively than the attention mechanism and feature fusion module. Compared with the Mask R-CNN and Mask Scoring R-CNN, the proposed method increased the segmentation precision by approximately 21% and 19.59%, respectively. These results demonstrate that the proposed method performs favorably against the other two methods of pixel-level aircraft skin defect detection.     

结构光三维视觉系统研究

介绍了结构光三维视觉原理,在广义坐标系中采用投影和透视变换理论建立了结构光三维视觉模型。在此基础上,提出了基于非线性最小二乘法的模型参数标定方法,并建立了结构光三维视觉系统,给出了实验结果。     

图像序列中机动目标三维运动和结构的计算

综合视觉运动分析中的2类处理方法,选取图像中的角点作为特征点,在理论上证明了图像序列的光流场可以近似地用角点的位移场代替。利用已有文献中的建模思想,详细推导出递归计算机动目标三维运动和结构的非线性计算模型,采用广义卡尔曼滤波(EKF)递归地计算图像序列中机动目标的三维运动和结构。合成图像序列和真实图像序列实验结果表明该算法能取得较好的效果。     

基于RBF神经网络的结构光三维视觉检测方法

研究了基于RBF(Redial Basis Function)神经网络的结构光三维视觉检测方法.该方法利用RBF网络良好的非线性映射能力以及学习、泛化能力,通过所获取的高精度的样本数据来训练RBF网络,最终建立起了用于结构光三维视觉检测的RBF网络模型.与常规方法相比,该方法不需要考虑视觉模型误差、光学调整误差等因素对视觉检测系统测量精度的影响,因而能够有效的克服常规建模方法的不足,保证了检测系统具有较高的精度.     

基于曲率的指尖检测方法

指尖检测是基于视觉的徒手人机交互系统的关键环节,由于背景的复杂性和系统的实时性要求,导致指尖的精确定位在处理速度和准确性方面存在很大问题。本文针对这一问题,提出了一种简单高效的基于曲率的指尖检测方法。该方法首先将输入视频流基于肤色空间进行二值化,并将二值化后的视频序列作为输入数据;然后采用边缘检测算法提取出肤色区域的边缘(肤色区域的轮廓),在肤色区域的轮廓上根据曲率来对类指尖的点进行检测,并且根据类指尖点的位置关系来判定一个肤色区域是不是手;最后通过过滤算法过滤掉误判手臂点。实验结果表明,该方法在不同的应用背景下都具有很好的检测效果,对光照的鲁棒性也较高,并且能够达到实时检测的效果。     

面向单目立体视觉的立方镜图像处理技术

面向单目视觉辅助光电准直仪对立方镜进行准直的需求,研究针对立方镜图像的数字图像处理技术。采用Canny边缘检测方法得到灰度图像中的边缘信息,借助Hough变换按照直线程度的强弱对边缘信息进行分类提取,得到图像中立方镜棱边对应的直线信息。按照偏角与截距对棱边直线信息进行分类排序,结合先验知识得到立方镜角点对应的配对相交直线,对配对直线求交点得到立方镜角点在图像中的位置,得到的立方镜棱边及角点信息为后续的立方镜位姿计算提供了基础数据。     

基于边缘信息的灰度目标跟踪算法

提出了一种基于边缘信息的跟踪算法,其可以实现对剧烈变化的灰度目标的精确跟踪.首先,利用基于双同心圆窗口算子的非线性边缘检测算法得到高质量的边缘信息;其次,为了解决单一边缘特征空间不能充分表征目标的难题,提出了一种通过组合边缘图像构建特征空间的方法,以便为目标建模提供充分信息;再次,在构建的特征空间中使用核估计方法对目标进行建模;在目标定位阶段,利用Kalman滤波器对目标进行预估后,再由Mean Shift算法在预估位置邻近区域实现目标定位;最后,提出了一种基于形态学的动态模型更新策略,使得算法不仅可以获得精确的目标区域,还可以实现对目标尺寸和形状变化的自适应.实验结果表明,本算法不仅可以有效跟踪剧烈变化的灰度目标,而且跟踪窗口可以实现对目标尺寸和形状的自适应.      

基于视觉模型和图像特征的遥感图像压缩

遥感图像具有纹理复杂、边缘丰富的特点,通常难以实现高压缩比.统计了图像经过小波变换后在不同方向、不同级别子频带的能量分布.然后基于人类视觉系统(HVS)模型中的对比敏感度函数(CSF)和遥感图像的能量分布特征,提出一种不同频带小波系数的变步长量化方法,并把该方法应用于嵌入式小波图像编码器.实验结果表明,与不采用心理视觉量化的压缩方法相比,在相同压缩比下,峰值信噪比略为降低,但恢复图像的主观质量有一定程度的改善.