Researches On The Alignment Technology Of Large Aperture Telescopes In Real Time

Due to the effects of gravity,wind loads,vibration and temperature gradient,the relative positions of the mirrors are misaligned,which can degrade the image quality of telescope systems.At present,the telescope alignment is one of the most important research topics.To simplify the complexity of measurement systems and improve the speed of calculating processes,the sharpness function method combining with Stochas-tic Parallel Gradient Descent(SPGD)algorithm is employed.However,for telescopes with large apertures and wide fields of view,the images are sometimes under-sampled for the finite pixel sizes and the restriction of applications,causing the aliasing effect and losing some information.As a result,the images cannot express the aberrations of optical systems accurately,decreasing the precision of the telescope alignment.In this paper,the telescope is aligned based on the sharpness functions of images with different sampling factors,and the main contents are as follows.Firstly,the vector aberration theory is used to study the characteristics of aberration distributions when the telescope is in designed and misaligned working conditions.The relationships between aberrations and misalignments are analyzed.Secondly,the influences which can affect the convergence and convergence speed of the telescope alignment method are analyzed.The results show that the parameters have to be chosen properly to ensure that the alignment program can converge quickly and stably.In this paper,reasonable ranges of the parameters are given for the simulation optical system.Finally,both simulations and experiments are carried out based on the sharpness functions of under-sampled and well-sampled images combining with the SPGD algo-rithm.In simulations,three kinds of sharpness functions are used,including the root mean square radius(RMS),absolute errors of the root mean square radius(RMS error)and relative errors of the root mean square radius(relative RMS error).The results indi-cate that using relative RMS error as the sharpness function can improve the alignment accuracy comparing with other two merit functions.Also,to ensure the image quality in the full field of view,a two-step alignment method is proposed.The first step is on-axis field alignment,and the second step is multi-field alignment.For multi-field alignment,the images corresponding to different fields of view have to be acquired by the same de-tector.Image targets have to be extracted to analyze the aberrations of each field of view.So in this paper,a new image segmentation method is presented to avoid the algorithm falling into the local optimum.The results show that both Gray-level Contrast(GC)and Uniformity Measure(UM)of the segmented images are better than 0.9,indicating that the images are well segmented.What’s more,the intensity distribution and image shapes are best preserved.Then,a valid target extraction and field matching method is presented and verified using ideal images and noise images.The results show that for both ideal and noise images,the Normalized Cross Correlation(NCC)is better than 0.9,indicating that all the targets are well extracted and the field matching is completed accurately.Then,the experiments are set up and the telescope is aligned according to the simulations.The correction results show that for well-sampled image whose Full Width at Half Maximum(FWHM)is 2 pixels,the average wavefront error correspond-ing to different fields of view after correction is 0.0548/1,and for under-sampled image whose Full Width at Half Maximum(FWHM)is 0.25 pixels,the average wavefront er-ror is 0.0726λ.Compared with the alignment precision of the well-sampled image,the precision of under-sampled image becomes worse,but the correction results still satisfy the requirement of image quality.