A whole-celestial star map recognition method

Abstract

The invention relates to an all-day star chart identification method. The method comprises the following steps: firstly, two star points A and B with the longest distance are extracted in a captured star map image; a star point C with the largest distance sum from the star points A and B is extracted from the other star points in the star map image; a star point D with the largest distance sum from A, B and C is extracted; star map areas matched with the star points A, B, C and D in distance and energy are searched from a guide star catalog 1 to serve as primary selection areas; if a to-be-identified star and image boundary meet a certain range in the star points extracted from the captured star map, three stars are taken to form a distance and angle combination in several ranges around the star respectively, and the distance and angle combination is used for identifying the star; after all the star points in the photographed star maps are identified sequentially, the star closest to the other stars is selected from the identified star points, and the stars outside the range of angle T are eliminated with the star as the center; remaining sets of identification stars are saved, andthe identification process ends.

Description

technical field [0001] The invention relates to a star map recognition method, which belongs to the technical field of star sensor design. Background technique [0002] In the process of developing the stellar gyro sensor, the precision requirement is high, and the number of stars in the star catalog is large. Using the original all-celestial navigation star recognition method, the recognition speed is slow, and it is difficult to quickly identify the given attitude of the stars. Therefore, for the large navigation star catalog, it is necessary to develop a fast all-celestial recognition algorithm. Based on the advantages of polygon recognition algorithm and grid recognition algorithm, this study proposes a quadrilateral grid method to identify stars in the whole celestial sphere for the actual situation that there are many navigation stars. The test results show that the algorithm can complete the global star recognition within 2 seconds on the DSP platform, with a recogni...

AI Technical Summary

Problems solved by technology

However, the training of these algorithms requires a large amount of calculation and requires a large training set to complete multiple pattern recognition, so it has the disadvantage of long recognition time; the accuracy is easily affected by the size of the training set and the length of training time; potentially requires a large The amount of storage is used to store weights, so the hardware requirements are relatively high

Images