| With the rapid development of intelligent building industry,fine agriculture,aerial survey and remote sensing,medicine and other industries,new requirements are put forward for the research and analysis of large-scale,refined and complex weak flow field.Background Oriented Schlieren(BOS)is a new visual display and quantitative analysis method of flow field.It has the advantages of non-contact,low cost,simple layout and can calculate the quantitative numerical information of flow field.However,the measurement accuracy of BOS method is limited by the sensitivity of the measurement system,the selection of background pattern and the calculation accuracy of image analysis algorithm.Many threshold parameters depend on artificial experience lack of applicability,and there is a lack of space-time analysis method of flow field structure.Therefore,in view of the above problems,the main research contents include:(1)This paper expounds the basic principle of background schlieren method,and introduces the measuring device and process of background schlieren method for targets with weak flow field.The influence mechanism of background pattern parameters is analyzed.Aiming at the weak flow field,the target background pattern with black background and white spots,speckle target background pattern with size and spacing of 2-5 pixels;The sensitivity influence parameters of the instrument layout model of BOS method are analyzed.For the object with weak flow field,taking into account the model sensitivity,imaging range and algorithm robustness,with 35mm lens and the sensitivity optimal layout of the imaging model of 1/2L_B≤L_D≤2/3L_B is determined.(2)Aiming at the difficulty of determining the fixed window threshold of cross-correlation algorithm,unifying the fixed window size,analyzing the problems of poor displacement accuracy and many false matches,an adaptive window cross-correlation algorithm constrained by image information entropy is proposed.Firstly,using the Mini Batch theory,a small number of search windows are randomly selected to calculate the image information entropy and calculate the mean value to determine the basic threshold;Secondly,the information entropy of all search windows is calculated iteratively and the iterative constraints are set.The window size is adjusted adaptively through the information entropy threshold constraints.Finally,the Gaussian nine point surface model is used to restrict the correctness of the results and eliminate the wrong matching results.The simulation results show that the adaptive window cross-correlation algorithm constrained by image information entropy is better than the optical flow algorithm and the original algorithm.In the speckle background experiment,the RMSE is only0.017 pixels,which is 19.05%higher than the original algorithm,and has good extraction accuracy and time efficiency.(3)For the fine extraction of flow field,the calculation of sub-pixel peak is studied.Based on the correlation matrix function model of background schlieren image,the sub-pixel fitting technology of correlation peak of two-dimensional Gaussian surface is simplified,and the three-point twice sub-pixel fitting method of one-dimensional Gaussian curve is used to determine the accurate peak.At the same time,in view of the lack of Time-Space Analysis Method in BOS analysis,this paper proposes a time-series baseline accumulation analysis method to ensure the accuracy of weak flow field structure extraction and detailed analysis.Finally,this paper extracts the weak flow field of flame,the weak flow field of blower and the very weak flow field structure of UAV.The experimental results show that the fine extraction method of weak flow field structure proposed in this paper successfully extracts the boundary conditions,numerical information and change law of flow field structure,and verifies the reliability and effectiveness of the method. |
PDF Full Text Request