This paper addresses the complete visual coverage path planning problem in raster terrain and presents an effective solution with hierarchy based on multiple observers siting and an evolutionary path planning algorithm. The main idea of the method is to get a nearly optimal complete visual coverage path by connecting all of the observers in the visual coverage set. Firstly a new deterministic algorithm for a quasi minimal number set of observers is presented, which is superior to the general greedy algorithm in the time complexity and the quality of solutions. On the basis of that we further present a hierarchy path planning method whose core is an evolutionary algorithm. The advantage of the proposed method is that it realizes really complete visual coverage with a comparatively optimal path and easy to extend to the version of multiple paths.