Increasing The Accuracy of Earth Horizon Sensor using Kalman Filter

Sensors for establishing satellite angle are used to keep the satellite balance in the space. One type of these sensors is Earth horizon image sensor which detect the image of horizon via a two dimensional detector. Through processing this image the horizon location in the image is determined. Comparison between the current situation with the desired situation of the horizon at balance attitude of satellite can provide the angles of the satellite. Information of angles are sent to controllers, and based on angles the proper signal are sent to actuators from the controllers, thereby the satellite would be maintained at the desired angle. Receiving the accurate and valid information about the Earth horizon sensor is essential, and increasing the accuracy of the sensor will enhance the angle control of the satellite. In the infrared images taken from the horizon, due to atmosphere of the Earth, the accurate location of horizon cannot be distinguished. Some general methods for determination of object edges have been reported; however, it is expected that using specific methods for detection of the Earth edge in image would result in more accurate information. On the other hand, use of model will help to distinguish the satellite angles. In the present study, a method is developed to determine the location of the Earth horizon. In this method, the curve of the Earth horizon is determined via information fusing of pixels around the horizon. Then after, using the correlation of this curve, the satellite angles will be calculated. Alternative to image processing method, the angles are also predicted via satellite model. Finally, results provided by image processing and results predicted by satellite model are combined through Kalman filter to calculate the accurate angles of the satellite. Results of implementation through the developed method showed that the method can accurately determine the Earth horizon and satellite angles. Keywords - Satellite, Earth Horizon Sensor, Image Processing, Pixels Fusion, Kalman Filter