Research On Image Registration And Repair Method Of Pressure Sensitive Paint

The pressure sensitive paint technology is a wind tunnel pressure measurement frontier technology with high economical efficiency and fast reaction speed.Compared with the traditional punching measurement,it effectively reduces the cost of model design and manufacturing and improves production efficiency.The principle is that the pressure-sensitive paint will perform dynamic oxidation quenching reaction according to different pressures,emit light of different intensity,and use the pressure-light intensity conversion formula to obtain the pressure.However,in actual experiments,due to the influence of strong winds,the model often undergoes deformation and displacement,especially nonlinear elastic deformation,and the acquired image model is also prone to distortion.The pressure calculation of the distorted wind image and the windless image will seriously affect the accuracy,so the image needs to be registered.Similarly,due to strong winds and other factors,the paint surface of the model often appears to fall off,so that the resulting image has holes,which will affect the final pressure test,so it is necessary to repair the holes of the pressure sensitive paint image.In response to the above questions,the main work and contributions of this paper are as follows:(1)This paper innovatively applies non-rigid point cloud registration technology to the field of pressure sensitive paint image registration.The non-rigid registration algorithm is more suitable for the case of nonlinear elastic deformation of the model,and the point cloud method is used to make the image detail area more effective registration,and it is also beneficial for the subsequent 3D reconstruction.(2)This paper proposes an algorithm with higher registration accuracy based on the characteristics of pressure sensitive paint.The algorithm transitions from coarse registration to fine registration.The coarse registration takes into account the pixel gray factor and finds a more favorable initial position.Since the two-dimensional non-rigid ICP algorithm only considers the two-dimensional coordinate positional relationship,the correlation of the pixel gray of the pressure-sensitive paint image is neglected,so that the registration accuracy is not high.However,misregistration occurs again by directly using the three-dimensional non-rigid ICP algorithm.Therefore,in order to further improve the registration accuracy,this paper proposes a non-rigid ICP algorithm based on pixel-based search strategy.The algorithm designs a two-objective search strategy that takes into account 2D coordinates and pixel gray values,and achieves accurate local matching point search.Optimized with dual goals.Finally,the algorithm of this paper was compared with five kinds of registration algorithms on multiple sets of pressure sensitive paint images.The experimental results show that the proposed algorithm has the best registration accuracy.Compared with the suboptimal algorithm,the RMSE is improved by more than 15%,and the NMI is improved by about 5%.(3)This paper applies the deep learning algorithm to the field of pressure sensitive paint image inpainting.Based on the reference literature and the specificity of the pressure sensitive paint image,a pressure-sensitive paint image inpainting algorithm based on deep learning was proposed.Because the pressure sensitive paint image has a special texture,its repair needs to meet the requirements of industrial precision and automation.However,the way in which image features are manually extracted is difficult to implement.Therefore,this paper selects the deep learning image inpainting algorithm based on texture information to repair the pressure sensitive paint image.At the same time,in order to adapt to the changes of the subsequent test environment,this paper has enhanced the generalization ability of the network.Finally,using objective evaluation indicators on multiple sets of data,this paper compares several classic image inpainting algorithms.The algorithm is better than other algorithms,which realizes the repair of pressure sensitive paint images and lays a foundation for pressure measurement.