Research On Star Identification Algorithm For Remote Sensing Tools

With remote sensing technology becomes more rapid development, technicalrequirements for satellite image positioning accuracy is gradually improving, the use ofstellar scaling is a new geometric calibration method. This paper focuses on the staridentification algorithms using stellar as a geometric calibration method and a newalgorithm named quadrilateral algorithm is proposed, which was applied to complete staridentification and visual axis scaling of a large FOV imaging deviceAccording to the way that we do ZXY rotation transformation matrix, this paperdeduces the rotation transformation matrix from the celestial horizon coordinate systemto the coordinate system of the star sensor, at the same time, deduces the translationaltransformation matrix from the coordinate system of the star sensor to the coordinatesystem of the image plane, providing theoretical foundation for star identification.Compared the advantages and disadvantages of several classical star identificationalgorithms, this paper proposes an improved-triangle star identification algorithm–quadrangle star identification algorithm.To complete the quadrangle star identification algorithm, the gray thresholdsegmentation algorithm and one improved connective region search algorithm are usedduring the image pre-processing of the star map, extracting information of the top20to30brightest star targets from the observed star map, and the centroid locating algorithmwith a threshold is used to extract the centroid location of these star targets in thecoordinate system of the image plane, and a observed quadrilateral database is built with6star angular distances between any two star targets of the four, stored in ascending order.Meanwhile, a star magnitude threshold algorithm is used to filter out the518stars whosestar magnitude value is under4.0from the standard star catalog， built and stored32232navigation quadrilaterals.In this paper, quadrilateral algorithm divided the star identification process into twoparts: quadrilateral matching and final star matching. Quadrilateral algorithm is used inthe star identification of actual star image taken by large FOV camera on earth underdifferent environmental condition, the simulation results show that Quadrilateral algorithm increases the dimension to6star angular distances, reducing the redundancymatch rate, and the success ratio is over90%, which shows that using Quadrilateralalgorithm in Geometrical calibration of large FOV camera is feasible, the repeatabilityprecision can be up to±0.04.