The Research Of Image Quality Improvement Of Synthetic Aperture Laser Radar

Synthetic aperture laser radar (SAL), whose image resolution ratio could beyond thediffraction limit, is now enjoying a rapid development in China and foreign countries.The record of the highest resolution ratio in azimuth direction refreshes frequently. Atthe same time, the bandwidth and the slope of the chirped emission signal of SAL cometo halt because of the limitation of material performance used in the signal radiationsystem. Thus, the resolution ratio in range direction undermines the image quality ofSAL seriously and makes SAL at a disadvantage. In order to ameliorate the imagequality of SAL, this thesis proposes a new scanning geometry of SAL, Zigzag scanning,which could improve the resolution ratio and therefore enhance the recognizability ofSAL target.The research significance of SAL Zigzag scanning is marked at first. The researchbackground and the hot research area of SAL are introduced in two aspects ofexperiment and algorithm. In the second Chapter, basic theory of SAL and its signalprocessing, also several widely used algorithms are explained. The idea of Zigzagscanning method, what is critical in this thesis, is stated. At the mean time, the systemrequirement and operation mode of Zigzag scanning is proposed. Assuming that theradar target is the Lambert reflector under physical optics approximation, the imagingsimulation of SAL is realized by MATLAB. The image registration of two related radarimage of SAL Zigzag scanning over the same target in different angle is achieved. Therotation and zoom information of the two radar image are obtained by using the methodof Fourier-Mellin Transform, and the phase correlation method was employed tomeasure translation between them. To conduct image fusion of the two related SALimage after the process of registration, six methods are adopted, including data addition,data multiplication, choosing the max value of each pixel, Laplacian Pyramid transform,SIDWT with Haar and Morphological Pyramid transform. Afterward, four parameters,RMSE, PSNR, Cross Entropy and correlation coefficient, are selected to describe theimage fusion quality. The result table which covers two original image and six fusionimage, indicates that in an ideal situation, data addition, data multiplication, choosingthe max value of each pixel have little effect on the improvement of SAL image quality.In contrast, Laplacian Pyramid transform, SIDWT with Haar and MorphologicalPyramid transform could improve the SAL image quality significantly, especiallyMorphological Pyramid transform method. Furthermore, resolution ratio of the SALZigzag scanning image is a very important parameter to judge the performance ofZigzag scanning. After imaging analysis of dot matrix target, compared with originalSAL image over the same target, there is a solid proof that SAL Zigzag scanning could improve SAL image resolution ratio and provide high resolution ratio in both twodimensions of radar image, which is the same as the resolution ratio of the originalazimuth direction. Finally, under the effect of atmospheric turbulence or3dB whiteGaussian noise, the image processing process of SAL Zigzag scanning is repeatedrespectively. The result shows that, under the influence of atmospheric turbulence or3dB white Gaussian noise, Laplacian Pyramid transform, SIDWT with Haar andMorphological Pyramid transform still have obvious amelioration of the image quality.In conclusion, the Zigzag scanning method is reliable and practicable in improving theSAL image quality and resolution ratio.