This dissertation investigates distribution density table based algorithms for attitude determination. Attitude specifies the orientation of the body frame with respect to a given reference system and can be determined by finding various parameters which denote relationship between body vectors and corresponding reference vectors. There are two main issues in the vector based attitude determination, one is a vector observation method and the other is an attitude estimation method from vector observation.
The first part of this dissertation is about vector observations using a star sensor. An alternative approach to vector observations for attitude determination using a star sensor is presented. While conventional star trackers require star vector observations through an identification of star constellations, the proposed method takes multiple vector observations of virtual line-of-sights instead of stars. A virtual line-of-sight is the pointing direction of small portion of a star sensor\`s field-of-view and is measured by searching celestial positions having the same theoretical star densities with the measurement, defined as the number of detected stars in the field-of-view. The suggested approach is based on the fact that the distribution of stars in the sky is not homogeneous. A mathematical measurement model is derived and distribution density tables made from star catalog are given as a reference frame. Since star streaks are countable even in the case of fast spinning body frame, the proposed method allows not only a stabilized spacecraft but also a rotational one to take measurements, in which latter case the vector observation of stars is usually difficult owing to the low signal to noise ratio of star streaks. The second part of dissertation deals with attitude estimation methods using distribution density tables. Algorithms suggested in this dissertation minimize the cost function defined as a weighted difference between the measurement seque...