Reconstruction of Smoke based on Light Field Consistency

Abstract

This paper proposes a method to reconstruct smoke by estimating its spatial density. In this method, light field cameras are employed to estimate an initial spatial density by generating refocused images. First, an initial spatial density is determined using the minimum density value in the estimated space of each light field camera. The initial spatial density includes a spatial blur caused by the aperture of the light field cameras. The spatial density is reconstructed by removing the spatial blur using a blur model. We have previously proposed the fundamental model for the method. However, the previous model had two problems. The first problem was that the model did not consider directivity of light scattering. The second was that the model assumed that the blur spread uniformly. Furthermore, the previously conducted actual experiment was only evaluated visually, which was insufficient. In this paper, we propose a method that can resolve these problems. Our method is numerically evaluated by an actual experiment. Our experimental results show that smoke and fog can be reconstructed using two light field cameras.