摘要
通过聚类算法仅能从二维图像中提取出教学要素信息,无法从多维空间提取要素信息,提取教学视觉要素存在精度差的弊端,设计激光多维成像教学视觉要素提取系统,采用激光视觉传感器获取不同教学视觉要素的三维激光图像后,通过信号处理电路对三维激光图像实施特征计算,提取出有价值的教学视觉要素;硬件控制电路实现激光光源、激光传感器和图像增强器控制参数与DSP间的数据双向通信;系统通过角点检测法获取粗糙的教学视觉要素,运用激光多维图像要素点与摄像机参数相标定方法,得到要素的精确坐标,实现激光多维成像教学视觉要素的精确提取。实验结果说明,所设计系统的教学视觉要素提取精度均高于95%,提取复数标准色和单色标准色+辅助色的用时分别低于0. 72 s和0. 55 s,是一种高精度、高效率的教学视觉要素提取系统。
The clustering algorithm can only extract the teaching element information from the two-dimensional image,can not extract the element information from the multidimensional space,extract the disadvantages of the teaching visual element with poor accuracy,and design the laser multidimensional imaging teaching visual element extraction system.After obtaining 3 D laser images with different teaching visual elements using laser visual sensor,the three dimensional laser images are calculated by signal processing circuit,and valuable teaching visual elements are extracted.The hardware control circuit realizes the two-way communication between the control parameters of laser light source,laser sensor and image Enhancer and DSP.The rough teaching visual elements were obtained by corner detection,and the precise coordinates of the elements were obtained by using the method of phase calibration of laser multi-dimensional image elements and camera parameters.The experimental results show that the precision of the visual elements in the design system is higher than 95 %,and the time used to extract complex standard colors and monochrome standard colors+auxiliary colors is lower than 0. 72 s and 0. 55 s,respectively.It is a high-precision and high-efficiency teaching visual elements extraction system.
引文
[1]罗爱民,施鼎.基于聚类分析的C4ISR系统服务关键要素自动提取方法[J].系统工程与电子技术,2016,38(5):1081-1085.
[2]岳建锋,杜博宇,王天琪,等.基于激光单目视觉系统的焊缝提取[J].天津工业大学学报,2015,34(6):84-88.
[3]侯佳,何志平,舒嵘.离轴三反光纤阵列激光三维成像发射系统[J].红外与激光工程,2016,45(4):117-121.
[4]薛成志,李亚敏,李贵叶,等.激光扫描光学断层成像系统的设计与应用[J].光子学报,2017,46(7):189-197.
[5] PAREKH V,JACOBS M A.SU-F-R-05:Multidimensional Imaging Radiomics-Geodesics:A Novel Manifold Learning Based Automatic Feature Extraction Method for Diagnostic Prediction in Multiparametric Imaging[J].Medical Physics,2016,43(6):3373-3374.
[6]段良君,祝连庆,娄小平,等.激光扫描测头视觉系统结构参数优化[J].应用光学,2016,37(2):308-314.
[7]佟军民,刘俊伯,胡松.基于塔尔博特自成像光刻机的照明系统设计与分析[J].激光与光电子学进展,2017,54(8):260-268.
[8]王秀平,白瑞林.双队列控制的激光视觉焊缝跟踪系统研究[J].计算机工程与应用,2015,51(2):30-34.
[9]张毅,杜凡宇,罗元,等.一种融合激光和深度视觉传感器的SLAM地图创建方法[J].计算机应用研究,2016,33(10):2970-2972.
[10]陈苏婷,胡海锋,张闯.基于激光散斑成像的零件表面粗糙度建模[J].物理学报,2015,64(23):105-113.
[11]黄敏松,雷恒池,陈家田,等.基于光电二极管阵列的成像测量系统设计及其应用[J].量子电子学报,2015,32(5):620-626.
[12] TAHARA T,QUAN X,OTANI R,et al.Digital holography and its multidimensional imaging applications:a review:[J].Microscopy,2018,67(2):55-67.
[13]叶琳琳,王晓曼,刘鹏.基于RMA的前视激光合成孔径雷达成像算法[J].半导体光电,2015,36(2):331-334.
[14]李博阳,耿楠,张志毅.基于机器视觉的激光条纹扫描系统[J].计算机仿真,2015,32(6):241-246.
[15]江伦,胡源,安岩,等.星载激光目标指示器光学系统研究[J].激光与红外,2016,46(5):542-546.
[16]田振国,张立,朱开毅.基于微片Nd:YVO激光回馈的共焦层析成像系统[J].光学精密工程,2015,23(10z):29-34.
[17]张合新,王强,张腾飞,等.激光主动成像图像边缘检测算法研究[J].电光与控制,2015,46(4):1192-1196.
[18] RUSSAKOVSKY O,DENG J,SU H,et al.ImageNet Large Scale Visual Recognition Challenge[J].International Journal of Computer Vision,2015,115(3):211-252.