An efficient task clustering strategy and an improved ant colony optimization algorithm are proposed for the the earth observation scheduling problem with a single agile satellite. Firstly, taking the field of view of the satellite into consideration, the task clustering strategy based on minimum clique partition in graph theory is presented and a series of cliques are built, which can enhance the observation efficiency effectively. Then, this paper presents an improved ant colony optimization algorithm to build the optimal observation path under a time window constraint and the constraint of attitude maneuver ability of the satellite. A novel optimal index containing the priorities of targets and the energy consumption for attitude maneuver is proposed to improve the energy efficiency. Besides, in order to overcome the shortcomings of the basic ant colony algorithm which is easily trapped into the region of local minimum, this paper designs a heuristic ant colony algorithm which synthesizes the priorities , time windows of targets and the transition time of the satellite between targets. Finally, a series of targets on the earth are selected and the effectiveness and efficiency of the algorithm proposed are demonstrated.