Depth estimation using Stereo Light Field Camera

Light field imaging has emerged as a new modality, enabling to capture the angular and spatial information of a scene. This additional angular information is used to estimate the depth of a 3-D scene. The continuum of virtual view-points in light field data efficiently handles occlusion and provides a robust depth estimate for smaller distances. However, a narrow baseline in a light field camera limits the depth estimation for larger distances. To have an efficient occlusion handling and increase the operating distances, we proposed a novel disparity based stereo light field depth estimation method. First, segments are obtained in central sub-aperture of left view and then estimate the disparity vector of these segments using left camera sub-aperture images. This handles occlusion efficiently. Then stereo disparity at boundaries of these segments exploiting the epipolar geometry inherent in a light field data. Finally this stereo disparity at boundaries is propagated to other pixels and normalized. We provided a synthetic stereo light field data-set having inherent characteristic of a light field. We have tested our approach on a variety of real-world scenes captured with Lytro Illum camera and also on synthetic images. The proposed method outperforms several state-of-the-art algorithms.