Multiple View Image Denoising
MetadataShow full item record
A pinhole camera (large depth of field) image capture is essential in many computer vision applications such as Simultaneous Localization and Mapping, 3D reconstruction, video surveillance. For these applications obtaining a set of clean images(less noise, less blur) is important. In this thesis, we propose a new approach to acquiring pinhole images using many pinhole cameras. The cameras can be distributed spatially to monitor a common scene, or compactly assembled as a camera array. Each camera uses a small aperture and short exposure to ensure minimal optical defocus and motion blur. Under such camera settings, the incoming light is very weak and the images are extremely noisy. We cast pinhole imaging as a denoising problem and seek to restore all the pinhole images by jointly removing noise in different viewpoints. Our Multi-view denoising can be used as a prior to the applications mentioned above. Our algorithm takes noisy images taken from different viewpoints as input and groups similar patches in the input images using depth estimation. We model intensity-dependent noise in low- light conditions and use the principal component analysis and tensor analysis to remove such noise. The dimensionalities for both PCA and tensor analysis are automatically computed in a way that is adaptive to the complexity of image structures in the patches. Our method is based on a probabilistic formulation that marginalizes depth maps as hidden variables and therefore does not require perfect depth estimation. We validate our algorithm on both synthetic and real images with different content. Our algorithm compares favorably against several state-of-the-art denoising algorithms.