Image Recovery for Electrical Capacitance Tomography Based on Low-Rank Decomposition

Images reconstructed by an existing image reconstruction algorithm for electrical capacitance tomography (ECT) are usually blurred or corrupted due to data noise. To recover the missing or corrupted pixels in the images, an approach to robustly reconstruct images based on low-rank decomposition is presented for ECT. With this approach, recovering the permittivity distribution is recast as one of recovering low-rank matrix from corrupted images. A convex optimization technique is used to obtain the correct low-rank matrix and the error matrix. Experimental results show the effectiveness of the proposed method. Furthermore, the simplified optimization method is fast and could be used for online imaging. The method can also be used in other tomographic techniques.