GPU (graphics processing unit)-based quick star catalogue retrieving method

Abstract

The invention provides a GPU (graphics processing unit)-based quick star catalogue retrieving method. The method comprises the following steps: (1) dividing a star catalogue into constellations to store by adopting a longitude and latitude partitioning method; (2) calculating the constellation covered by a round view field area of a star sensor, wherein spherical coordinates of a visual axis are calculated at first in the process, boundaries of longitude and latitude of the round view field are calculated, and the constellation covered by the round view field is searched in the partitioned star catalogue; (3) transmitting stars in the constellation into the global memory of the GPU to store in a star set mode, dividing a star operation thread grid of the GPU into a plurality of thread blocks, wherein each thread block is responsible for processing a star set, and each thread is responsible for retrieving a star; and (4) retrieving stars falling in the round view field by each thread, and retrieving stars falling in a charge coupled device (CCD) imaging plane in the round view field. By the method, the retrieving time in each view field is controlled in the millisecond level, so that the real time requirement of constellation generation and other simulation applications to star retrieving in a large view field and wide star magnitude domain can be fulfilled.

Description

technical field [0001] The invention belongs to the field of computer simulation, and in particular relates to a star catalog retrieval method with fast real-time characteristics. The retrieval method can quickly retrieve stars falling into the visual field in the star catalog under a given field of view. Background technique [0002] The star sensor is an important on-board device in the satellite attitude determination and control system. It mainly uses a fast star map recognition algorithm to determine the satellite attitude. In order to test the recognition algorithm performance of the star sensor, it is necessary to use a star simulator to simulate the star map in the field of view of the star sensor at any time and at any attitude of the aircraft. The computer simulates the background star map captured by space equipment such as star sensors. First, the field of view is determined with the given size of the star-sensitive field of view and the rotation angle of the fie...

AI Technical Summary

Problems solved by technology

In order to speed up star catalog retrieval, literature [1] (refer to Bone Jeffery W. on-orbit star processing using multi-startrackers [J]. Proc. SPIE, 1994, 22(21): 6-8) proposed a spherical matrix structure to store the star catalog, and improve the efficiency of star access by optimizing the star catalog structure, but this method cannot significantly reduce the access time when the matrix dimension is large

Literature [2] (referring to Rao Caijie, Fang Jiancheng. A method for extracting observed stars in star map simulation [J]. Optical Precision Engineering, 2004, 12(2): 129-134.) proposed the division of longitude and latitude circles Star catalog, use right ascension and declination circles to divide the celestial sphere into non-overlapping areas, and store each area independently. This method has clear boundaries and simple calculations. However, when dealing with large-scale star catalogs, it is easy to generate too many partitions, which will affect the stars. Query efficiency; literature [3] (refer to Zhang Guangjun, Wei Xinguo, Jiang Jie. An improved recognition method for triangular star charts [J]. Acta Aeronautical Sinica, 2006, 27(6): 1150-1154.) By constructing the "inner Connect the cube" to evenly divide the celestial sphere into 6 blocks, then subdivide the blocks into more sub-blocks, and read the stars through the sub-block index numbers. This method is convenient for querying, but it is more complicated to judge the block boundaries of stars. When the value is large, the boundary calculation takes a long time

In order to narrow the search scope, literature [4] (refer to Hu Yining, Gong Yan. Design and implementation of dynamic star map display algorithm [J]. Acta Astronautics Sinica, 2008, 29(3): 849-853.) field, take the eight sub-areas adjacent to the partition where the optical axis of the field of view is located as the search area, and retrieve the stars falling into the field of view. The length of the longitude arc gradually shortens as the latitude increases. When the boresight points to different sky areas, the number of sub-areas covered by different fields of view is not the same.

Literature [5] (Refer to Cai Zhiwu, Han Chunhao, Chen Jinping. A method for constructing a high-precision spaceborne navigation star library [J]. Journal of Surveying and Mapping Science and Technology, 2006, 23(1): 29-32.) through calculation The latitude and longitude boundary range of the field of view on the spherical surface, and then find the sub-area associated with the field of view, but this method uses the expansion factor to calculate the latitude and longitude boundary range, and the calculation accuracy is not high

In addition, the existing star retrieval process in the field of view is implemented serially on the CPU, and the retrieval efficiency is not high when the number of stars is huge

Images