Reseach On Spaceborne SAR Simulation And Image Processing Technology

Synthetic Aperture Radar(SAR)constructs an equivalent antenna through the movement of the radar and uses pulse compression technology to acquire two-dimensional high-resolution radar images in both range and azimuth direction.After 70 years of SAR development,different carrier platforms and various working modes have emerged.The application of SAR is widely applied in the fields of military and civilian.Considering the cost and efficiency,the spaceborne SAR has played an important role in various fields.Besides,the extensive research and application of SAR has increased the demand for high-quality SAR images.In addition to the improvement of imaging algorithms,a wide variety of SAR image processing technique have become more and more important.This thesis starts with the basic principles of SAR and the particularity of satellite trajectories.Then the working mode and signal generation of spaceborne SAR are analysed.The BP algorithm are adopted to process multi-operating mode spaceborne SAR echo signals.And the imaging results influenced by non-ideal factors is studied.Moreover,the radiation correction,geographic coding and SAR image filtering get well studied.The main work and innovation of this thesis are as follows:1.The basic geometric relationship of SAR and the definitions of important terms are researched,such as azimuth,oblique view and zero Doppler plane.The special properties of SAR signals are analysed.The range and azimuth resolutions of SAR are derived and the approach of obtaining the high-resolution SAR images is analyzed.2.The characteristics of trajectory and various working modes in spaceborne SAR are analyzed.The relationship between the aiming point of the spaceborne SAR simulation and its calculation process are analyzed.The generation process of the spaceborne SAR echo signal is analyzed,and the pulse compression technology is applied to achieve the high resolution of the SAR image.The characteristics of SAR imaging algorithms in time and frequency domains are compared,and the advantages of Range-Azimuth domain BP algorithm in spaceborne SAR signal processing are analyzed.The imaging performance of BP imaging results of multiple working modes on surface targets and point targets was analyzed,and the index evaluation was performed.According to the effects of simulation imaging,the advantages and disadvantages of the working modes in SAR image performance were summarized.Through simulation experiments,the effects of different degrees of non-ideal factors on the imaging quality of spaceborne SAR are analyzed.3.The methods of SAR image processing are studied.The first is to address the problem of uneven SAR image energy caused by electromagnetic waves radiated by the antenna.The method of antenna azimuth correction to compensate for the uneven radiant energy generated by the antenna in different radiation directions is researched.In view of the different radiant energy intensity of the target at different oblique distances in the antenna and the imaging scene,performed distance compensation according to the radar radiation equation.The theory of relative radiation calibration factor is analyzed and its calculation process is deduced,which can make the target of the same RCS have the same pixel value on the SAR image.Through the performance analysis of radiation correction,the radiation accuracy of the SAR image is improved after the radiation correction,and the pixel values of the same RCS target are more uniform in the SAR image.The second category is geographic information coding of SAR images.After calculating the projection range,the SAR image is projected from the range-azimuth domain to the WGS84 coordinate system by the projection algorithm.The errors in geocoding are analyzed,and it is verified that the accuracy of geocoding is between single-digit pixels.Finally,the application of the DnCNN method is proposed to reduce multiplicative SAR speckle noise.Compared with traditional filtering methods,the filtering network based on DnCNN has significant advantages both in image detail restoration and speckle denoising.