Research On Coordinate Detection Of Observed Images In Astronomical Positioning And Orientation

In astronomical orientation and orientation observation,celestial body sensors and photographic theodolites are often used to observe celestial bodies so as to obtain celestial body image information for orientation and orientation.For the observation target image pixel coordinates need to be extracted to provide the raw data for the subsequent positioning and orientation astronomy,so the coordinates of the observation target image extraction accuracy directly affects the positioning accuracy of astronomical orientation.In this paper,mainly focuses on representative observational targets,such as:the star map,the sun image,and the coordinate of the reticle of the theodolite obtained by specular reflection.The main work of the study is as follows:(1)First of all,the paper carries out image preprocessing on the star map,including denoising of star map filtering,threshold segmentation of star maps,and division of connected regions.Then the target star point sub-pixel subdivision extraction is performed on the processed star map.The simulated star maps were simulated by the traditional centroid method,weighted square centroid method,thresholded centroid method,Gaussian surface fitting method and skewed normal distribution PSF correlation method.Experimental results show that the accuracy of the target's center-of-mass coordinates based on the skewed normal distribution PSF correlation method proposed in this paper can reach around 0.05pixels,and the accuracy is higher than that of the other four algorithms,which provides a guarantee for the follow-up of high-precision astronomical orientation and orientation resolution and other important part.(2)Then the paper deals with the observed sun image processing.Firstly,three kinds of threshold segmentation methods are introduced to segment the image of the sun image.It is concluded that the OTSU method is suitable for extracting the sun image.Then the algorithm edge detection combined with Sobel operator,Spatial moment operator is proposed to detect the edges of the sun image.Then for the degraded sun image,this paper proposes an effective circle edge detection method.Finally,the paper proposes three kinds of solar image centroid extraction algorithms according to the principle of least squares,and compares the accuracy of the three algorithms to extract the centroid(geometric center)coordinates of the solar image.(3)Finally,the crosshair center coordinates are extracted from the crosshair images obtained by specular reflection by a theodolite,and an algorithm combining Hough transform and least squares is proposed.First,a series of image processing work is performed on the obtained cross hair image.Then the processed crosshair image is first fitted with Hough transform and least squares to fit the linear equation,and the coordinates of the intersection point are calculated as the central coordinates of the required crosshair.The experimental results show that the accuracy of the algorithm proposed in this paper is within O.Olpixels when there is no noise interference,and the absolute deviation of the coordinates extracted by the Gaussian noise algorithm is within 0.2 pixels.