| Streak tube lidar is a new type of three-dimensional imaging lidar based on image converter scanning measurement technology,which has great demand in military and aerospace fields.The development of streak tube lidar is the demand of national strategic high-tech field.Traditional lidar is difficult to recognize targets with a certain depth of field and a small contrast.The streak tube lidar has a large coverage area of each laser pulse,and its search efficiency is much higher than that of non-imaging lidar.It is an instantaneous imaging system that can move both the detector and the target at the same time.Targets can be effectively identified under the cover of the jungle and night.Based on the hardware design of streak tube lidar system and the research of target three-reconstruction algorithm,the fast three-dimensional reconstruction algorithm,the geometric distortion correction of streak image and the improvement of compressed sensing imaging quality are studied in this paper,taking the measured target detection data of lidar as the research object.The main work includes the following aspects:Firstly,streak tube camera is the core equipment of lidar imaging system.A special picosecond visible streak phase for lidar is designed and manufactured.Explore the application of new electronic components to design the scanning circuit,and test the parameters of the scanning circuit using the circuit test platform.Engineering application shows that the designed circuit is safe and reliable.The static and dynamic experimental platforms of the streak camera are built to detect and calibrate the parameters of the streak camera under the static and dynamic working modes.The spatial modulation function,the ultimate spatial resolution and the ultimate spatial resolution of the full screen of the fringe camera with different off-axis distances under static and dynamic modes are analyzed.In order to construct the lidar imaging system,the preliminary work should be done well.A fast algorithm of object three-dimensional reconstruction based on sparse sampling and image interpolation is proposed.The validity of the fast algorithm of object three-dimensional reconstruction is verified by analyzing the relationship between sampling rate and time of object three-dimensional reconstruction and similarity of object three-dimensional reconstruction matrix.When the sampling rate is 4%,the sampling bilinear interpolation method is 2.5 times faster than the original reconstruction method,the correlation coefficient of the three-dimensional reconstruction matrix is 0.93,the sampling nearest neighbor interpolation method is 3.4 times faster than the original reconstruction method,the correlation coefficient of the three-dimensional reconstruction matrix is 0.92,and the detection distance is 706 meters for the target.The speed of sampling bilinear interpolation method is about 3.3 times faster than the original reconstruction method,and the correlation coefficient of the three-dimensional reconstruction matrix is 0.99.The reconstruction speed of sampling nearest neighbor interpolation method is 4.5 times faster than the original reconstruction method,and the correlation coefficient of the three-dimensional reconstruction matrix is 0.98.With the decrease of sampling rate,the reconstruction time decreases,and the reconstructed streak image frame is reconstructed.The more the number of reconstructed images,the faster the reconstructed images are,and the better the reconstructed images are.Next,the aberration equation of the object points of the image converter is analyzed.Because the magnification of the object points with different off-axis distances is different,the image obtained on the object surface fluorescent screen is distorted.The image distortion caused by the relative change of the radial magnification is changed into pillow distortion.The correction model and concentricity of the spherical optical system are analyzed.The distortion model of spherical imaging system is obtained,and two mathematical models are matched.The experimental results show that the geometric radial distortion rate of the tested image is reduced from 14.7% to 1.23%,the edge brightness(gray value)of the image is increased by 28.2%,The experimental results show that the proposed method can correct the imaging distortion of the streak camera and improve the contrast of the image.The above-mentioned distortion correction algorithm is used to process the fringe image of the close-range target of lidar,and the calculated value of the target distance before and after correction is compared.The accuracy of the target distance calculated by the corrected fringe image is improved.A method of streak image denoising based on the effect of three-dimensional reconstruction of target is proposed.By analyzing the noise characteristics of lidar imaging system,the streak image denoising is not to pursue subjective evaluation,but to extract peak features.Experiments on streak image denoising with various filtering algorithms show that the total variation filtering algorithm has good denoising effect and the mean filtering denoising efficiency is high.Compressed sensing is proposed to reconstruct the target steak image of lidar.Before reconstructing,a semi-threshold processing is carried out,that is,the gray value is set at 0 below the threshold value and the gray value is kept at the original value when it is greater than the threshold value.A total variation regularization compression sensing reconstruction algorithm is adopted.The experimental results show that the average contrast of the streak image is increased by about 3%.The proposed method is used to process the target fringe image of lidar,select the appropriate threshold,compare the calculated values of target distance before and after correction,and the relative error of target distance calculation after correction is reduced. |
PDF Full Text Request