节点文献
物体表面轮廓光学动态检测与特征识别
Research of Object Surface Profile Optical Dynamic Measurement and Identification by Characteristic
【作者】 付林;
【作者基本信息】 南京理工大学 , 光学工程, 2007, 博士
【摘要】 光学三维轮廓测量由于其非接触性、高精度与高分辨率、测量速度快、自动化程度高,被公认是最有前途的三维轮廓测量方法,物体三维表面轮廓测量在现代工业及实际生产中起着越来越重要的作用,对动态物体与目标的三维轮廓特征的快速检测与识别尚存在高精度动态过程检测的问题,这已成为目标特征识别、工业与工程在线检测的热点和亟待解决的问题。在广泛调研、归纳和总结的基础上,对国内外研究物体表面三维形貌测量这一学科领域内容和发展前景有了一个全面的了解。再根据我们的条件,本文主要针对动态目标的三维轮廓的检测和特征识别问题,在激光测距扫描特征轮廓提取与结构光条纹三维轮廓检测与识别开展研究。(1)提出了利用最小二乘估计器构造自适应滤波器的思想,基于目标本身形态的总体尺度范围特征设定滤波窗口,结合目标先验特征,建立了激光测距扫描特征轮廓提取与目标识别定位模型,实现从三维动态物体的一维特征数据进行目标定位识别的目的。通过算法分析与实验结果分析表明,该算法能够满足实时条件下的目标识别要求。(2)提出了利用自适应标准模版匹配,分选破损的公路路面,解决高速公路破损路面结构光图像分选问题,从而实现对高速公路路面形变的快速实时检测。(3)提出一种数字三色条纹相移形貌投影栅线法,通过单帧条纹投影图像,利用三步光强相移算法完成物体表面的形貌重建;提高了从单帧条纹投影图像重建物体三维轮廓的质量,不需要相移装置而误差较小,成本较低,简易实用,将有利于推广到动态实时分析。
【Abstract】 Optical measurement of objects’ 3D surface profiles is well-known as themost promising 3D profile measurement method for its features such asnon-contact measurement, high accuracy and high resolution, high speedmeasurement and high automation degree. The measurement of objects’ 3Dsurface profiles is more and more important in the modern industry andpractical manufacture. But high precision and dynamic procedure measurementfor rapid measurement and identification of dynamic objects’ 3D profilesand characteristics becomes the hotspots and the desiderating issues ofobject charactertistic identification and industry and engine onlinemeasurement. Based on wide-ranging investigation and generalization,according to our condition, the researching direction had been determinedto study of characteristic distillation from laser distance measurementprofile and 3D profile measurement of stuctrued light fringe for themeasurement and characteristic identification of dynamic object’s 3Dprofile.1. This paper introduces a new arithmetic approach to the objectidentification of lidar range image by using the least squares estimationand linear glide updater for the design of a filter. The filter window isset up in accordance to the characteristics of object shape size and objects’prior characteristic. This arithmetic approach can have an easy access tothe process of object identification. The arithemetic analysis andexpirements results illuminate that the arithemetic is fit to real-timeobject identification.2. An algorithm of self-adopted standard template matching is introduced.The algorithm, which based on characteristic image identification, pick outthe distress pavement road surfaces, solve the structured light imagesclassification problem of highway pavement distree road surface and realizehigh-speed and real-time measurement of high-way pavement surfacesdeformation.3. The tricolor fringe phase-shifting technique in digital projectiongrating profilometrY to reconstruct the profiles of object surface is proposed. Only one fringe pattern is used to reconstuct the profiles of objectsurface by the advanced thchnique combined with a three-step phase-shiftingalgorithm without any phase-shifitng devices. The technique improves theobjects’ 3D profile reconstruction quality from single fringe pattern byprojection grating. The proposed technique is economical and practicle andmaybe use for the dynamic tea-time analysis. The experiment illuminates thatthe technique is right and feasible.
【Key words】 Profile construction; projection grating profilometry; Fourier transform; phase-shifting algorithm; lidar; object identification;