Depth Estimation from Defocus Images Based on Oriented Heat-Flows

The amount of blur on the defocus image depends on the depth information of the scene. So depth of the scene can be estimated by calculating the blur with the knowledge of the lens parameters. A novel depth estimation mode based on oriented heat-flows is proposed in this paper. In this model, the process of image defocusing is described using oriented heat-flows diffusion. The diffusion can be seen as the coupling of two weighted heat flows along orthonormal directions. The diffusion directions are defined by the local coherence geometry of image and the diffusion strength is the function of blur amount. Experimental results show that the model is quite effective and the emergence of artificial depth information in edge can be avoided by using this model.