Research On Micro-displacement Measurement System Based On Monocular Image Sequence To Estimating Clear Imaging Position

There have been many kinds of detection methods for small displacement,and the accuracy of measurement has been continuously improved.High-resolution micro-displacement measurement technology mainly includes non-optical measurement technology such as electricity,microscope and other measurement methods and optical measurement technology represented by laser interferometer.Due to the continuous expansion of the field of industrial measurement and the continuous improvement of measurement accuracy,the classical contact measurement can no longer meet the requirements of the industry,while non-contact measurement has many advantages and has become a research hotspot in the field of measurement.In this paper,through monocular vision and laser point combination in non-contact measurement,theoretical research and experimental verification of high precision measurement of small displacement are carried out.The characteristics of pinhole imaging and lens imaging are introduced,and a micro-displacement measurement model based on lens imaging principle for camera bias and laser source tilt is proposed.The influence of various parameters in the model on the resolution of the measurement system is analyzed.The principle that the central coordinate value of the imaging spot changes significantly when the depth information changes slightly at the initial position(y)of the model is expounded.This shows that the measurement system has high resolution to the micro-displacement of the object in the neighborhood and can improve the measurement accuracyBy analyzing the influence of various parameters in the measurement model on the measurement accuracy,several important parameters of the measurement system are determined,such as focal length,light source offset distance,light source inclination angle and initial position.According to the determined parameters,a measuring device is designed,manufactured and installed.The measuring device comprises a camera offset adjusting mechanism,a light source tilting adjusting mechanism,a lens adjusting mechanism and a target moving mechanism.Based on VS2010 and OpenCV,a prototype system for micro-displacement measurement using monocular vision and laser spot combination is developed,which includes the functions of reading images,image processing,obtaining spot center coordinate values,controlling linear motor movement,denoising algorithm,clear imaging position estimation,measuring displacement display,etc.From the theoretical model of measurement,it can be found that the central coordinate value(y)of the imaging spot has great influence on the measurement accuracy,and ensuring the accuracy of the central coordinate value of the spot is one of the key research contents in this paper.Therefore,by analyzing the reasons for the formation of background noise and edge noise in speckle images,an algorithm for removing background and contour noise at fixed points and an algorithm for removing background and contour noise based on statistical principle with outward expansion and inward contraction and double ellipses are proposed.Both algorithms can effectively remove background noise and contour noise,and the search efficiency of the latter is better than that of the former.Experiments show that the stability and accuracy of the proposed two algorithms are better than the classical gravity center method and Gaussian method.On the basis of obtaining the center coordinate value of the speckle image by using the double ellipse de-noising algorithm of outward expansion and inward contraction,and aiming at the problem that the speckle image is blurred due to the change of depth information and a clear speckle image is obtained again,a method for estimating the optimal position of the lens based on the speckle image sequence is proposed.The method estimates the position of the clear image by using the short path of the speckle image sequence,and calculates the center coordinate value of the clear image corresponding to the position.The theoretical data and experimental results verify the consistency of the proposed estimation algorithm and analytical method,and the estimated clear imaging has the largest pixel mean.Through experimental research,when the imaging spot is in a clear position and the lens moves for a small displacement,the definition of the imaging spot is almost unchanged,and the coordinate value of the corresponding spot center changes little and is difficult to distinguish.Further analysis shows that the focal point of the lens is not an ideal point,but a tiny area,which leads to the above phenomenon.In this case,if the target has a small displacement,the central coordinate value of the imaging spot changes obviously.In view of this,this paper divides the measurement range into five sub-regions,each sub-region estimates the clear imaging position according to the image sequence,and then uses the moving target to obtain different central coordinate values of the spot at the estimated position to calibrate the measurement model.In this paper,three calibration methods are proposed,i.e.fitting algorithm based on theoretical measurement model,micro-line segment and cubic polynomial.Through experimental verification and comparison,the calibration method based on micro-line segment has higher accuracy but complex model.The calibration method based on cubic polynomial has simple model but low measurement accuracy.However,the calibration method based on the theoretical measurement model is simple and has high precision.Its relative error is less than 0.2%(the maximum absolute error within the measurement range of 5 mm is 8um),which can realize high precision measurement of small displacement.