Research On Squint And Mylti-angle Image Registration And Height Estimation

Synthetic Apertuer Radar(SAR) is a microwave imaging radar system which can operate the targets under the all-day and all-weather condition.Due to the limition of traditional linear SAR's observation angle,we can only obtain the limited information and we can't perform three-dimensional imaging. The range of multi-angle SAR's observation is much larger than the traditional syetems,and it can acquire high-resolution SAR images and have three-dimensional imaging capability.The multi?angle SAR system have wide application in military investigation,disaster monitoring and other fields.Multi-angle SAR is becoming a new hotspot in radar research with its unique system advantage.Geometric rectification is one of the key technologies of image preprocessing,which is a premise for high precision positioning and recognition of SAR images.In SAR image constellation model,the target height is an important parameter for locating the target geographic coordinates,So the target height resolution and target positioning and geometric correction are coupled together.Multi-angle SAR can achieve height estimation based on different squint angles,in order to gain high-precision three-dimensional information,the main work of this paper are as follow:(1)Summarize the unique advantages of multi-angle SAR system.We analyze the change of the SAR parameters with the oblique angle ,and analyze the constellation model in case of squint,we deduce the square points' Doppler center and the corresponding satellite time formula.(2)We analyze the influence factors of the three-dimensional measurement accuracy.Then,we analyse the errors and the problems caused by the resampling method through the simulation experiment.Aiming at the problem of more complex geometric distortion caused by squint SAR,we propose a resampling method which can preserve relative position. The experimental results show that the method can effectively improve the geometric correction of squint SAR.(3) Using the MSTAR dataset,we use SIFT vector matching and RANSAC elimination mismatch strategy.Then,we match the images in a certain range of view, obtain the coordinates of the same name point.(4) Using the MSTAR dataset,we use the position offset and the angle difference in the image of the same name point at different angle to get the height of target.The experiments show that the result of the measurement is close to the true value.