Effective seeding method is the key to influence the streamline distribution and to understand the underlying properties of flow field. Based on the accurate description of flow field variation and important features, this paper proposes two information entropybased seeding methods to solve the wellknown occlusion and cluttering issue. The first greedy seeding method locates interesting areas through the calculation of entropy values. The greedy seeding method is highly sensitive to the important features. The second Monte Carlo seeding method generates random inputs based on a probability distribution, and then defines the influence areas of input grid points as a circle in 2D and a sphere in 3D. Comprehensive experiments on multiple datasets show that the greedy seeding method can capture the important features efficiently and the Monte Carlo seeding method shows significant ability to obtain global variation. Besides, the combination of both methods can get more optimal flow field visualization.