本文選題：攝像機標定 + 軸徑��； 參考：《吉林大學》2017年碩士論文

【摘要】：在精密測試技術領域,機器視覺測量技術是一項新興技術,該技術可以實現(xiàn)零件的尺寸測量、目標的視覺跟蹤及機械裝置的運動分析等功能。機器視覺測量技術利用計算機代替“人眼”,看到圖像并理解圖像,對攝像機采集的數(shù)字圖像信息特征進行提取,最終完成視覺檢測。軸類零件是各類機械裝置中應用最廣泛的一種零件,隨著技術的探索與進步,對機械裝置的精密度以及運轉(zhuǎn)速度的要求越來越高,在機加工過程中常常需要對軸類零件進行高精度的加工,因此對軸類零件直徑尺寸的準確測量顯得尤為重要。本文根據(jù)軸類零件的自身幾何特征及光學成像過程,提出一種軸徑測量方法。首先,本文根據(jù)小孔成像原理建立攝像機的成像模型,分析了攝像機標定技術�；趶堈训钠矫鏄硕ǚ�,在畸變模型中考慮兩項徑向畸變和兩項切向畸變,以此修正光學鏡頭畸變,進而提高了攝像機內(nèi)部參數(shù)和外部參數(shù)的標定精度;為提高攝像機參數(shù)標定精度,通過攝像機標定實驗,確定合適的角點檢測數(shù)目以及背光源光照強度。然后,為提高圖像測量的檢測精度,從圖像中獲取被測物體的亞像素邊緣位置是至關重要的。本文首先理論分析了幾類常見的亞像素邊緣檢測方法,包括灰度矩法、空間矩法、Zernike正交矩法、高斯擬合法、雙曲正切擬合法、梯度插值法以及三點插值法;最后以實際測量尺寸的誤差作為亞像素邊緣檢測精度的評價準則,通過測量實驗,確定了穩(wěn)定性較好并且檢測精度較高的梯度插值法,作為本文軸徑測量的亞像素邊緣檢測方法。接下來,本文首先根據(jù)軸零件在攝像機下的成像過程,建立軸徑測量的數(shù)學模型,并著重減小了軸線與圖像平面不平行所帶來的誤差;然后對軸徑測量數(shù)學模型中的未知參數(shù)進行標定;最后,考慮到測量軸徑的實用性與便捷性,提出了一種基于目標軸的真實軸徑與像素軸徑比值的標定系數(shù)方法。最后,通過軸徑測量實驗,驗證本文提出的軸徑測量方法的精度及標定系數(shù)方法的實用性。對兩種軸徑測量方法的測量結(jié)果進行比較,并分析了影響測量精度的誤差因素,為提高測量精度確定方向。對于機器視覺測量技術的發(fā)展來說,本文的研究工作有一定的工程應用意義。
[Abstract]:In the field of precision measurement technology, machine vision measurement technology is a new technology, which can realize the functions of measuring the dimension of parts, tracking the vision of the target and analyzing the motion of the mechanical device. The machine vision measurement technology uses the computer instead of the "human eye", sees the image and understands the image, extracts the information feature of the digital image collected by the camera, and finally completes the visual detection. Shaft parts are one of the most widely used parts in all kinds of mechanical devices. With the exploration and progress of technology, the precision and running speed of mechanical devices are becoming more and more demanding. In the process of machining, it is often necessary to process the shaft parts with high precision, so it is very important to measure the diameter of shaft parts accurately. Based on the geometrical characteristics of shaft parts and optical imaging process, a method for measuring axis diameter is presented. Firstly, according to the principle of pinhole imaging, the camera imaging model is established and the camera calibration technology is analyzed. Based on the plane calibration method proposed by Zhang Zhengyou, two radial and two tangential distortions are considered in the distortion model to correct the distortion of the optical lens, and the calibration accuracy of the internal and external parameters of the camera is improved. In order to improve the precision of camera parameter calibration, the appropriate number of corner detection points and the illumination intensity of backlight are determined by camera calibration experiments. Then, in order to improve the accuracy of image measurement, it is very important to obtain the sub-pixel edge position from the image. In this paper, several common sub-pixel edge detection methods, including gray moment method, spatial moment method, Gao Si fitting method, hyperbolic tangent fitting method, gradient interpolation method and three-point interpolation method, are theoretically analyzed. Finally, the error of actual measurement dimension is taken as the evaluation criterion of sub-pixel edge detection accuracy. Through the measurement experiment, the gradient interpolation method with good stability and high detection accuracy is determined. As a subpixel edge detection method for axis diameter measurement in this paper. Then, according to the imaging process of the axis parts under the camera, the mathematical model of the axis diameter measurement is established, and the error caused by the axis is not parallel to the image plane is reduced. Then, the unknown parameters in the mathematical model of axis diameter measurement are calibrated. Finally, considering the practicability and convenience of measuring axis diameter, a calibration coefficient method based on the ratio of real axis diameter to pixel diameter is proposed. Finally, the accuracy of the proposed method and the practicability of the calibration coefficient method are verified by the experimental results. The measurement results of the two methods are compared, and the error factors affecting the measurement accuracy are analyzed to determine the direction for improving the measurement accuracy. For the development of machine vision measurement technology, the research work in this paper has certain engineering application significance.
【學位授予單位】：吉林大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TH133.2;TP391.41

【相似文獻】

相關期刊論文 前10條

1 王社陽;宋申民;陳興林;強文義;;改進的空間矩亞像素邊緣檢測算法[J];哈爾濱工業(yè)大學學報;2006年06期

2 徐中宇;寧闖;周亞洲;;一種快速亞像素邊緣檢測算法[J];吉林大學學報(理學版);2014年02期

3 孫維健;;一種新的亞像素邊緣檢測算法[J];數(shù)字技術與應用;2014年02期

4 付鵬;高曉蓉;;基于矩的亞像素邊緣檢測算法的對比研究[J];微計算機信息;2007年18期

5 張虎;達飛鵬;李勤;邢德奎;;橢圓目標的亞像素邊緣定位方法研究[J];光學技術;2010年01期

6 林欣堂;王宗義;;一種新的快速通用二維圖像亞像素邊緣檢測算法[J];應用科技;2010年08期

7 詹林;劉秀珍;陳科澎;;亞像素邊緣定位在光纖中應用[J];數(shù)字技術與應用;2010年06期

8 艾澤潭;石庚辰;代俊;;微零件圖像亞像素邊緣定位算法[J];北京理工大學學報;2011年03期

9 李云;夏若安;;基于曲線擬和的亞像素邊緣檢測[J];重慶科技學院學報(自然科學版);2008年06期

10 田原Z，

本文編號：1958942