基于损失自注意力机制的立体匹配算法研究

针对现有的立体匹配算法在阴影、物体边缘和光照反射等区域匹配困难且存在大量错误结果的问题,设计了一种可拆卸的损失自注意力网络（loss self-attention net,LSAnet）查找图像中的匹配困难区域。LSAnet的网络各层相互稠密连接,应用了空洞卷积来扩大感受野,并以立体匹配算法生成的损失分布为标签,能够动态地进行有监督训练,最终生成匹配困难区域掩膜辅助立体匹配网络进行更好的优化;同时,改进了立体匹配网络中经典的特征匹配代价卷结构,降低了后续3D卷积的计算负荷,提高了匹配效率。实验结果表明,该算法相比于基准算法精度更高,并且可以提高算法对于匹配困难区域的鲁棒性。     

Aligning Education with Vision 2030 Using Augmented Reality

Vision 2030 requires a new generation of people with a wide variety of abilities, talents, and skills. The adoption of augmented reality (AR) and virtual reality is one possible way to align education with Vision 2030. Immersive technologies like AR are rapidly becoming powerful and versatile enough to be adopted in education to achieve this goal. Technologies such as AR could be beneficial tools to enhance maintainable growth in education. We reviewed the most recent studies in augmented reality to check its appropriateness in aligning with the educational goals of Vision 2030. First, the various definitions, terminologies, and technologies of AR are described briefly. Then, the specific characteristics and benefits of AR systems are determined. There may be a significance of the pedagogical method used by adapting the AR scheme and the consistency of the equipment and learning experiences. Therefore, three kinds of instructional methods that stress roles, location, and tasks were evaluated. The kind of learning that is offered by the distinct kinds of AR approaches is elaborated upon. The technological, pedagogical, learning problems experienced with AR are described. The potential solutions for a few of the issues experienced and the topics for subsequent research are presented in this article.     

A novel vision-based method for 3D profile extraction of wire harness in robotized assembly process

Automating stages for deformable objects in the production line, in which assembling a wire harness into a predefined position is a complex task owing to the specialized characteristics of the objects. Besides a few automatized systems proposed in the other studies to implement this task under simplified setup conditions, a significant portion of this process remains to be completed manually in industrial environments. To construct an automatic wire harness assembly system, the development of a method that can automatically detect the wire harness profile in a 3D environment and, consequently, guide robot arms to implement assembly tasks is indispensable. Therefore, this study presents an approach that satisfies this requirement, which not only proposes a deep learning-based system to detect the wire profile, but also improves the accuracy of the detected results through a correction method according to the depth values of contiguous areas. The verification of the approach in a robot system that highlights its usefulness and practicality demonstrates the potential of the proposed method to replace people and consequently, reduce labour costs in factory environments.     

深度学习的图像实例分割方法综述

实例分割是一项具有挑战性的任务,需要同时进行实例级和像素级的预测,在自动驾驶、视频分析、场景理解等方面应用广泛.近年来,基于深度学习的实例分割方法迅速发展,如两阶段检测器Faster R-CNN扩展出的聚焦于网络的精度而非速度的强大实例分割基准Mask R-CNN,一度成为实例分割的标杆.利用高速检测的单阶段检测器延伸出的实例分割算法YOLACT填补了实时实例分割模型的空白,具有较高的研究和应用价值.本文首先对实例分割算法进行了类别划分,然后对一些代表性的算法及其改进算法进行了深入分析,并阐述了相关算法的优缺点,最后对实例分割方法未来的发展进行了展望.     

基于Matlab实时运动目标跟踪检测系统

运动目标的检测跟踪是视频理解技术和计算机视觉的研究热点,其在解决智能视频监控,人机交互,智能交通系统等领域有着广泛而重要的应用,基于此利用matlab平台构建出USB摄像头实时图像采集处理系,从视频流采集到处理综合利用了背景估测,图像分割,目标检测与跟踪算法准确高效地检测出环境场景中的动态目标,并成功地对其进行实时追踪。利用matlab的simulink模块编程实现提取视频流YcbCr输入系统进行运算处理,并改进了背景估测和目标检测算法,提高系统的实时性。最终利用多次试验,对室内和室外运动目标实现检测跟踪,验证了系统处理实际问题的可靠性能。     

基于计算机视觉的油封尺寸检测系统设计

针对油封的大批量、多品种、流水线作业的生产特点以及"零废品率"的检测要求,利用机器视觉检测和计算机图象处理技术,研究开发了适应大批量生产和在线实时检测的油封尺寸机器视觉检测系统;建立了系统软硬件结构,分析了系统的组成和工作原理及工作过程,并对图像处理算法进行了设计;系统标定后的实验检测结果证明,该检测系统能实现产品100%检测的生产目标,具有非接触、在线实时、精度适中、成本低等特点,具有较好的生产通用性。     

Theory of edge perception for computer vision feedback control

In this paper, the authors address the problem of edge-perception for its applications to vision-feedback control in robotic systems.In natural vision, the recognition of objects takes place through the process consisting of eye system, neural networks and cognition. The cognitive process, in turn yields a phenomenon known as perception. This is the phenomenon of perception of physical attributes, such as edges, color and texture, etc., which is responsible for the recognition of objects through the natural vision processes.In this paper, we make an attempt to postulate the theory of perception for gray-level images. The gray-level images, when going through the cognitive and perception processes, are contaminated by the uncertainty; here we call it cognitive uncertainty.The studies in this paper are confined to the phenomenon of edge-perception for two-dimensional gray-level images, however, these studies can be extended to other types of visual attributes both in two-dimensional and three-dimensional spaces. Indeed, the perception of these attributes, which attempts to emulate the human vision system, may help in the design of a truly robust computer vision-feedback control system for robotic applications.     

激光焊接对接拼缝测量的微景深方法

对于复杂微细、错边量小的激光焊接拼缝测量,现有的视觉传感器方法不能稳定、可靠地提取出拼缝的中心位置、拼缝宽度等信息。为此,提出了基于微景深的激光焊接对接拼缝测量方法,该方法采用高倍光学放大镜头、工业CCD(Charge Coupled Device)/CMOS(Complementary Metal-Oxide-Semiconductor Transistor)相机和LED(Light-Emitting Diode)外部照明装置,以光学成像光轴与被测表面法矢呈45°的方式,组成的对接拼缝的非接触测量系统,既解决了显微放大时景深很小、测量范围小的问题,又解决了狭窄拼缝高光学放大倍数拼缝检测的难题。实验结果表明,该系统可快速准确地提取出拼缝中心坐标、拼缝宽度以及拼缝局部表面法矢。拼缝宽度测量精度优于5 μm,拼缝中心测量精度优于6 μm,可识别的最小拼缝宽度为0.02 mm,完全满足激光焊接对接狭窄拼缝的检测与自动跟踪的要求。     

基于视觉的多特征手势识别

手势是一种自然直观的交互方式,基于视觉的手势识别是实现新一代人机交互的关键技术。本文在已有的手势识别技术基础上,从手势分割及手势表示两方面着手,提出了一种单目视觉下的手势识别方法。利用颜色特征检测肤色区域,成功分割出人手;利用人手的轮廓及凸缺陷检测指尖,再利用指尖的数目和方位来表示一个手势,进而结合轮廓长度和面积等几何特征完成手势识别。传统的指尖检测方法需要遍历并扫描手掌外轮廓,计算量大,本文通过凸缺陷检测指尖,减少了计算量,提高了指尖检测的速度。实验结果表明,本文的方法具有很好的鲁棒性及实时性,能适应环境的变化。