Design of Fast Star Image Extraction Module for Autonomous Star Sensors Based on FPGA

Measuring attitudes of spacecraft accurately is required by the modern space explorations. Autonomous star sensors are one of the most advanced devices to measure attitudes as good as arc-seconds. In a measurement cycle, there are mainly four steps, i.e. star imaging via cameras, star image extraction, star pattern identification and attitude estimate among which star image extraction spends far longer time than the other steps since the star maps usually have very high resolution in order to achieve a very high precision. In this paper, the nearly real-time star image extraction module which utilized the chip of ZYNQ-XC7Z010-CLG400-1 as the core of the processor developed by Xilinx Inc. was designed based on FPGA when considering noise disturbances. First, the basic theory of the star image extraction was introduced. Then, the hardware structure was described. The program to process image maps and extract star images was designed. Star maps were processed by the global threshold segment method after the threshold of the intensity was calculated by OTSU algorithm. Positions of the star image centroids were calculated via connected domain analysis and the centroid calculation after the star map was processed by morphology processing. The procedures of the star image extraction were described in detail. Finally, the design results of the star image extraction module were presented. Experiments were fulfilled when simulated star maps with Gaussian noise were input into the module. It proved that the developed module of star image extraction extracted the star images correctly. As for the star maps with 1024×1024 pixels, the average processing time was less than 25ms. The position accuracy was superior to 1/18 pixels.