Research On Deflection Detection Method Of Simple Supported Beam Based On Machine Vision

Many large-scale industrial testing equipments adopt the structure of simple-supported beam.The deflection of simple-supported beam will occur due to long-term load and other reasons.The deflection will lead to low measurement accuracy of testing equipment.Therefore,the deflection detection of simple-supported beam is of great significance for correcting the measurement value of testing equipment and improving the measurement accuracy.When inspecting the inner surface quality of cylindrical members,a simple-supported beam structure is adopted.In this paper,the 10-meter-long simply-supported beam in the detection scheme is taken as the research object,According to the structural characteristics of simply supported beams,a deflection detection scheme based on machine vision is designed.A laser beam in horizontal direction is formed by installing a laser beam on the support of one end of the simple beam.A receiving screen is installed on the mobile trolley of the simple beam.The laser spot on the receiving screen is collected by a CCD camera to obtain the spot image.Through image segmentation and other processing,the position change of the spot in the detection process is obtained,and the deflection of each position of the simply supported beam is calculated.Firstly,this paper introduces the scheme of deflection detection,completes the hardware design and device selection,and then introduces the realization of software scheme.Aiming at the processing of spot image,through the comparison of mean filter,median filter and Gauss filter,the median filter is selected to remove the noise of the image.When the spot image is segmented,the methods of threshold segmentation based on maximum entropy,threshold segmentation based on iteration method and Otsu method are compared.Finally,a two-dimensional Otsu method based on gray level and neighborhood mean is adopted to improve the image.After threshold segmentation,the spot image with the same size is extracted,and the edge of the extracted spot image is optimized by morphological method.Because of the position relationship between the CCD camera and the light screen,trapezoidal aberration exists in the spot image.In this paper,a coordinate transformation method based on control points is adopted to correct the aberrated image.Through comparative analysis,bilinear interpolation method is selected to reconstruct the gray level of the image,and the checkerboard template is used to calibrate the image.The coordinates of the pixels and the reality are obtained.The correspondence of distance.Finally,experiments are carried out to verify that the centroid method,gray center of gravity method and circle fitting method are used to locate the spot center.Through comparative analysis of the data,the gray center of gravity method has good effect,and the deflection curve of a simple beam 10 meters long is measured.Experiments show that the accuracy of the deflection of simply supported beams can be less than 0.05 mm by using machine vision technology,which provides the accuracy guarantee for the high-precision detection equipment based on the structure of simply supported beams.