Contour Recovery Technology Based On Sinusoidal Grating

With the rapid development of frontier fields,such as aerospace,machining,medicine and biopharmaceutical,more and more three-dimensional measurement techniques are applied to them.The prospect of contour recovery technology is promising.With the progress of nuclear magnetic resonance technology,engine blade processing and repair technology,holographic projection and 3D printing technology,the usability,convenience and accuracy of traditional 3D contour recovery technology are beginning to be tested.Up to now,many foreign scholars have studied sonic wave measurement and electrical measurement sensing technology in the field of 3D measurement and used it.The domestic research is just limited to the contact 3D measurement stage.However,the contact compared with non-contact three-dimensional measurement type,has many disadvantages such as large cost,low accuracy.In view of this research status,this paper presents a photoelectric combination,based on the three-dimensional contour of the grating optical path recovery technology.The main innovations and accomplishments of this paper are as follows:(1)A self-tuning algorithm based on DLP projection system is proposed.In the structure design of the three-dimensional contour recovery system based on the optical path of the sinusoidal grating,this paper introduces two methods based on the digital projection system and the simulation structured light system.Considering the difficulty of the phase shift in the algorithm design and the distortion check in the projection image preprocessing,a digital projection system based on DLP and CCD is finally set up.The self-correction algorithm proposed in this paper is different from the traditional phase compensation mechanism.By adopting the phase mapping theory,the mapping relationship between the projection phase and the acquisition phase is calculated,and the distortion caused by electronic noise and the sinusoid is solved.(2)A new phase unwrapping technique is proposed which is different from the traditional method of branch expansion and phase unwrapping from line to line.The idea of this algorithm is taken from the flooding method in routing polling,which is the reflection of retroactive thinking.The proposed algorithm avoids the problem that the traditional method cannot accurately phase-expand feature points and edge points.Flooding method combined with edge extraction effectively divides the image into blocks and recursive expansion.Compared with the traditional phase unwrapping method,the accuracy has been greatly improved.(3)A set of fitting and forecasting techniques based on neural networks in depth learning theory is proposed.The technology implements an intelligent system,changing the method of point error correction from a traditional man-made manual to a non-objective reference.Neural network training is still under study,but the results of fitting and forecasting are not in violation of objective data and are reliable and feasible in comparison with the entity.The results will be in a more accurate result of a complete set of sinusoidal-based 3D contour restoration algorithms.