Celestial Navigation System (CNS), Inertial Navigation System (INS), Global Navigation Satellite System (GNSS), such
as GPS, GLONASS, GALILEO and Compass etc, and the integrations of them are some methods of autonomous
navigation for space. But these methods must be depended on the high speed links of the communications network.
Moreover, the precision of CNS is always worse, and can not meet the rigorous requirement of the space activities. INS
can not be used for long-term space navigation applications for its errors being accumulated. High accuracy can be met
by Carrier Differential Global Positioning System (CDGPS), but it is difficult to calculate the ambiguities of CDGPS.
Fortunately, autonomous relative navigation based on machine vision is a direction all over the world currently, and is
very suitable for autonomous spacecraft navigation because it has some advantages, such as low-cost, high precision,
autonomous ability, easy practicality etc. In this paper, on the basis of the attitude dynamics of spacecrafts and the theory
of machine vision, an autonomous relative navigation algorithm for spacecrafts based on dual quaternion and EKF is
proposed. The basis transform unit of this algorithm is feature line. Moreover, this algorithm is used to calculate both
relative position and attitude organically, and the disadvantages of those algorithms, in which, relative position and
attitude must be deal with separately, are overcome. Finally, the simulations show that this algorithm is an accurate valid
method for space navigation applications.