Large Size Object Deformation Detection Based On Binocular Stereo Vision

The deformation detection based on binocular stereo vision is a simple,nondestructive,real-time and non-contact deformation measurement method.This method overcomes the shortcoming of traditional deformation measurement method.It plays an important role in scientific research and practical engineering application.This paper focuses on the research of large size object deformation detection based on binocular stereo vision,the main contents are as follows:(1)Aiming at the problem of camera calibration in the deformation detection of binocular stereo vision,the calibration method used in this paper is based on the calibration of MATLAB and Zhang Zhengyou calibration based on OPENCV.Compared with the method of MATLAB toolbox calibration,Zhang's calibration method based on OpenCV is used to extract sub-pixel pixels from diagonal points,and the calibrated results are evaluated by the angle reverse projection to obtain the calibration average error of each image.The camera calibration is completed and the maximum likelihood estimation of the result is calculated.Comparing the calibration results of the above two methods,the advantage of the Zhang's calibration based on OpenCV is that the calibration speed is fast,and the calibration results can be directly used in the subsequent stereo correction and stereo matching.The advantage of the other one is that it can display the re-projection error,the 3D image,and calibrating every picture can be used different windows separately.(2)Aiming at the problem of image matching is proposed in this paper.The image matching based on point identification.A human identification point with distinct characteristics is pasted on the surface of the key part of the object.Using FC-IE camera software to control two CMOS cameras to capture image images synchronously,after preprocessing the image,the left image is used as reference image,and the Hough transform method is applied to detect the center and radius of the left image identification point.The image pixel coordinates of each identification Circle center are obtained.the size of the column pixels is first compared,and the row pixels are sorted in order.The right image do the same processing.If the image detected around the center coordinates of the same number,then is the point of matching..(3)Aiming at the problem of object deformation detection,this paper presents an object deformation detection method based on the identification point.After the image matching is finished,the three-dimensional coordinates of the matching points in the image are restored.The method is to restore the three-dimensional coordinates of the identification points after the left and right images are matched,compared the three-dimensional coordinates of the two states,and finally get the displacement deviation of the surface and the displacement deviation from the plane,then we can know whether the object is deformation?The distance between the measured object and the camera is adjusted to judge the effect of the field size on the deformation detection,and the experimental results show that the larger the error of the View field is.Finally,the strain displacement of the object is obtained by taking a large size object for deformation detection.The result shows that the deformation detection accuracy of large size objects based on binocular vision is good,and the efficiency of object deformation detection is improved effectively.