Research On The Extraction Accuracy Of Laser Stripe Center In Light Saturation And At The Edge Of The Object

The triangulation-based 3D scanning technology is widely used in the fields of agriculture,medical care,and high-speed rail trains.Its purpose is to acquire high-precision threedimensional point cloud data on the surface of an object,and the laser light strip center extraction is the key to scanning.However,in the actual three-dimensional scanning process,when light saturation occurs or laser scanning to the edge of the object,the extraction accuracy of the existing extraction algorithm is not high enough to meet the demand.Therefore,this dissertation aims to develop a high-precision light-bar center extraction algorithm for “three-dimensional scanning system of line-structured light based on binocular vision”in light saturation and scanning to the edge of an object.Thesis mainly completed the following work:(1)The method of image pre-processing was mainly studied.The median filter is used to remove the image noise according to the characteristics of image noise,and the OTSU is performed according to the image features.The pre-processing method is used to preprocess the light bar image when the light saturation or scanning to the edge of the object occurs.(2)An algorithm for accurately extracting the center of the light bar is proposed in light saturation.When light saturation occurs during scannig,the grayscale values of the pixels near the center point of the light bar are the same,and the pixels at other positions still conform to the Gaussian distribution.Then the complete stripe light intensity distribution is fitted,and the maximum point is taken as the center point of the light bar.The experimental results show that the improved algorithm has better extraction effect and higher precision than Gray gravity method,Geometric center method,Maximum value method in light saturation.The effect of light saturation to the extraction accuracy is avoided.(3)An algorithm for accurately extracting the center of the light bar is proposed when laser scans to the edge of the object.Scanning to the edge of the object,a part of the laser light falls on the background,resulting in the incompleteness of the light bar.First,the complete stripe light intensity distribution is fitted,and finding out the maximum value point.Then using the maximum value point compares with the relative positions of the pixels at the edges,to determine the final center point of the light bar.The experimental results show that the improved algorithm has better extraction effect and higher precision than Gray gravity method,Gaussian curve fitting method,Maximum value method at the edge of the object.The effect of the incompleteness of the light bar image to the extraction accuracy is avoided,and the scanning requirement to the extraction accuracy is satisfied.