A Rigid Visual-based Pose Measuring Method Applied Surface Reconstruction

Machine vision measurement technology has a widely application in the field ofmanufacturing, robot systems, diagnostically techniques of medically assisted visualnavigation, sophisticated detection technology, surface measurements andreconstruction. After a comprehensive review on the indoors and overseas currentsituation and developing tendency of machine vision measurement technology, apostural posture visual measurement method using for surface reconstruction ispresented in this subject. We regard the handheld measurement devices as a rigid bodyand research object. Based on machine vision principle and P4P problem, we establishhe the measuring mathematical model,The photographic image after processing, usingimproved algorithm based on least squares ellipse fitting and image-processed toextract the feature point coordinates and the internal and external parameters of thecamera. Based on rigid transformation theory, it can calculated center coordinates ofrigid body’s head and world coordinates of the point on the surface contact with probeaccording to the radius compensation principle. And this process can achieve surfacereconstruction. The two components of the measurement system, the flow chart ofeach module, and an improved algorithm based on least squares ellipse fitting aregiven in this article. A measurement system is developed by MATLAB.The main work and creative research in this article are as follows:1Design framework of the hardware and software are given from overallstructure, a measurement system is developed by MATLAB. Using histogramequalization estimated the threshold of image. It can preserve the edge as well asreducing the noise of images and controlling the accuracy of feature point edgeposition. Then the noise of image is smoothing filtered by Gaussian, select imagesegmentation with dual threshold method and extract the edge of feature elements.Lastly, extract accurately the gravity coordinates of feature point based on leastsquares fitting ellipse algorithm. 2The article studies on the camera calibration, contrasted with several commonlyused calibration method. In this paper, the camera is calibrated by three-dimensionaltarget, and calibration theory for a detailed study.3Introductive theoretically the principle of rigid body postural attitude Visualmeasurement. After analyzing the principle of the PNP problem the system’smathematical model is established. And calculation method of rigid body posturalattitude is given.4According building experimental platform, we can carry out the experiments ofcamera calibration and system measurement. The feasible degree of measurementmethod presented in this issue is proved from the practice. In this paper, the sevenvertex on the gage block as character points are selected, building the cameracalibration model. Because calibration board is of high manufacturing precision, theaccuracy of this calibration method can satisfy need. By means of comparing theexperimental results with precision instrument measured and analyzing the accuracy,speed and reliability of system. We can confirm the accuracy and stability ofequipment, finally proved the measurement method is feasible.