Research On Improvement Of High Resolution SAR Image Quality

Synthetic Aperture Radar(SAR) is a high resolution imaging radar working independently of solar illumination and under all-weather conditions, which has been widely used in all fields. But there are still image quality problems that limit its forward applications. In this thesis, SAR image contrast enhancement and sidelobe reduction methods for SAR image quality improvement are researched.SAR imaging model is discussed firstly, and two effective contrast measurements are studied, which are TEN and EME based on Webber theory. They are used to evaluate structure and dynamic features of SAR image in contrast enhancement. IRW, PSLR and ISLR are introduced to assess sidelobe level of SAR image. These assessment methods are used in the following researches of contrast enhancement and sidelobe reduction methods of SAR image.SAR image contrast enhancement methods are studied. Limitations of traditional contrast enhancement methods are discussed firstly. And then focus on the study of fuzzy enhancement method. A new fuzzy image contrast enhancement method based on adaptive crossover point is proposed. It analyzes the limits of the classical fuzzy enhancement method, and applies an adaptive crossover point and new enhancement operator to make it adaptive for SAR image by using the features of the image’s histogram. The performances of the method are assessed by a series of experiments.The formation mechanism of sidelobes in SAR image is discussed. Spatially variant Apodization(SVA) reduces sidelobes without degrading mainlobe resolution. But its weighting function has different peak values as parameter varies. In the thesis, an improved SVA method is proposed that uses a normalized weighting function and gives the optimal parameter. Experiments show that the new method reduces sidelobe better. In the discussion of sidelobe assessment, it is concluded that nonlinear sidelobe reduction methods broaden SAR image’s spectrum, resulting in inaccurate sidelobe assessment by using FFT interpolation. And the relationship between the degree of spectrum broadening and the data’s over sample rate is discussed.