Research On Mirror-like Detection Technology Based On Binocular Vision And Fringe Reflection

With the rapid development of industrial technology,industrial products using mirrors and mirror-like objects are becoming more and more diversified.At the same time,higher requirements are put forward for the processing and measurement accuracy of products.In many existing detection equipment,the image recording product is used to capture the object,which can be converted into image information for processing,only obtain two-dimensional information,but not the depth information.Therefore,the efficient and accurate 3-D topography measurement technology is becoming a research hotspot with the increasing demand of mirror-like object application.The binocular vision method has the advantages of simple structure and fast measuring speed,but there are characteristic points in the field of mirror-like object detection.The fringe reflection method uses the relationship between the fringe phase and the gradient to obtain threedimensional information of the object to be tested,which is often used to detect the surface of highly reflective objects.This paper mainly combines the advantages and disadvantages of the two methods,aiming at the binocular vision and fringe reflection technology to study the surface reconstruction of mirror-like objects and the quality detection of mirror-like surfaces.The main research contents include:(1)The principle of camera calibration and system calibration in binocular vision and fringe reflection measurement system is introduced in detail.Aiming at the situation that the camera can not directly collect the target of display screen,the coordinate transformation relationship between camera coordinate system and display coordinate system is determined by using camera calibration technology based on mirror reflection,and the calibration of binocular camera is realized by combining with MATLAB toolbox.(2)Introduced the principle of phase-shifting method to acquire the phase of object surface,introduced and analyzed various phase unwrapping technologies,and selected the path tracking algorithm based on quality chart to complete the unwrapping phase.This method can effectively suppress the linear overlap of phase errors.Compared with global unwrapping technology,this algorithm has shorter calculation time and higher efficiency.(3)The effective range of the detected object is determined by mask method,the surface normal vector is determined by polar constraint and surface normal vector uniqueness constraint,and the derived gradient data is calculated by Southwell integral reconstruction algorithm to obtain the depth information of the surface of the object to be measured.(4)An experimental platform based on binocular vision and fringe reflection system is built to measure the concave spherical reflector with high reflectivity.By comparing the fitting experimental data with the standard data,the reconstructing effect of the surface is analyzed,and the verification accuracy reaches the micron level.The application of fringe reflection method in surface defect detection of mirror-like objects is introduced,and two kinds of improved defect detection schemes are proposed.