Research On The Mechanism Of Star Image Motion Blur And The Methods Of Blurred Star Restoration Of The Star Sensor

Star sensor is beyond dispute the most accuracy sensor in the absolute attitude measurement, which is better than one arc second, and is basic and irreplaceable sensor in the celestial navigation. Therefore it could be utilized to calibrate other attitude measurement sensors or systems and is widely applied in the field of the navigation and attitude measurement. However, its accuracy could decrease when it is employed under the atmosphere as the various weather and dynamic conditions. Specifically, the star image will generate a blurred trajectory when its carrier is under maneuvering, which leads to the star image motion blur and increase the difficulty or even failure to the extraction of the star image. Under this circumstance, star pattern identification and attitude calculation cannot be conducted. This dissertation focuses on the mechanism of the star image motion blur and the restoration methods of the blurred image of the star sensor, which is based on the application of the vessel on the sea. The main works are as follows:1. Basic principles and advances of the star sensor are introduced and a deep survey of the conditions of the compensation of the star image motion blur is made, through which it concludes that all the compensation methods are trying to reduce the effect of the related motion between the object and the sensor.2. In order to research the mechanism and its origins, the features of the radiation of the star source are researched, and then the image spot distribution is given. The blur of the star image can be described by the degeneration function with the convolution of the source image. So the concept and characteristics of the convolution are discussed. Finally, the mathematical model of image blur is given and the conclusion is that the related motion and noises are the two main contributors to the blurred trajectory. The features of the carrier are also analyzed, and the results show that the blur mainly occurs under the rotation motion. Experimental results demonstrates that the earth rotation will influence the quality of the star image. The gradient direction of the star image is correspondent to this rotation but it needs no extra compensation as the extraction algorithm is limited. It is excited to find that we can also evaluate the extraction algorithm through this rotation. The exposure time is also explored and advices and principles of the selection of the camera and lens are provided under the consideration of reducing the effects of the motion.3. A simplified model of the star image motion blur is established, through which it is clear that the degree of the blur is proportional to the exterior parameters: exposure time, rotation velocity and the ratio between the focus length and the pixel size. Simulations and experiments have proved this model to be practical. But thorough analysis of its error components shows that the square of the tangent of the incident angle should not be neglected when the incident angle or the rotation velocity or both are big. As a result, the complicated model should be adopted.4. Image restoration principles using the deconvolution method is introduced. Simulation comparison between the traditional methods demonstrates that noise is a pivotal role to solve the restoration. Detailed research indicates that background and stray light is the major noise source.5. An inverse restoration method based on the correlation match template is proposed, simulation result manifests that it could improve the extraction accuracy by two to three times. A deep research in this method indicates that it will not stop work until the signal is totally covered by the noise.