Imaging Performance Improvements And Analysis Of Optical Synthetic Aperture System

Optical synthetic aperture system is expected to exceed the aperture limitation to achieve continuous high-resolution earth observation from geostationary orbit, except that the sparse pupils and phase errors restrict its imaging performance. In this dissertation, the methods to improve the imaging performance of optical synthetic aperture system was studied, which involved performance index design, pupil structure optimization, piston error detection, multi-frame imaging and restoration method, ringing elimination as well as pixel shift analysis. The main contents are as follows:(1) Imaging performance index construction. During the fill factor ranges, imaging performance of typical arrays were compared by features indexes and objective evaluation scales to reveal the relationship between image quality and spectral distribution. According to frequency value and spectral uniformity, the weighted frequency energy curve index was designed, which can accurately reflect the performance of synthetic aperture imaging system with any array structure or characteristic parameters, to provide design basis for imaging system research.(2) Pupil structure optimization. The principle to maximize middle and low spectral energy was proposed to optimize the array arrangement of three kind of six-aperture arrays and the centerclustered arrays were obtained. It had been approved that the new structures own better performances on imaging quality, restoring quality, error tolerance and registration precision than typical arrays, the imaging performance had been significantly improved.(3) Piston error detection. The quantitative indicators were designed to measure the random multiaperture piston errors and tip/tilt errors. Then the phase error tolerances to guarantee precise restoration were obtained. The conclusion that the relation between point spread function and piston error is irrational with coordinates along the perpendicular bisector of baseline had been deduced and proved. Therefore the effect of position accuracy to error detection is avoided. The distribution characteristic of far-field image on baseline determines the range of piston error and by which a unique value of piston error is obtained. Direct deconvolution and total variation denoising were used to eliminate turbulence and noise and to improve the detection accuracy.(4) Multi-frame imaging scheme design. Multi-frame images with various spectrum could be obtained by rotating the array and the result retored by multi-frame wiener filter could avoid the effect of phase errors. The optimal multi-frame imaging scheme was determined based on statistical experiments and the phase error tolerances of synthetic aperture system was improved. The method was extended to the one-dimensional non-redundant array. By increasing the rotation times, the spectrum was covered repeatedly so as to effectively compensate for the spectrum covering defect caused by the directional light intensity distribution. The availability of the method was validated and the effective range of fill factor was determined by experiments.(5) Ringing artifact detection and elimination. A perceptually relevant ringing artifact detection and removal method was proposed to conquer the influence of phase errors. The texture region was obtained by the scales of gradient direction randomness and local variance distribution. The whole edges and main edges were separately detected by Canny operator with different thresholds. The whole edges belong to the neighborhood of all edges were considered as the edge ringing candidates. The ringing region and flat region can be derived by combining the texture masking of human visual system. Based on the image classification, the adaptive bilateral filter with different parameters was applied to realize ringing artifact removal and edge textures preservation, the effects of the phase error on restored image quality were overcome.(6) Image pixels shift analysis. The influences of fill factor, array structure, imaging scheme, phase errors, image post-processing and other factors on the synthetic aperture image pixels shift were analyzed using feature points matching method. It turned out that center-clustered arrays, rotated imaging and multi-frame restoration, ringing artifact removal not only could optimize synthetic aperture array imaging performance, but also helped to reduce the image pixels shift.