Research On Electrical Impedance Imaging Method Based On Block Sparseness

Electrical impedance tomography(EIT)is a technique for reconstructing the conductivity distribution by injecting currents at the boundary of a subject and measuring the resulting changes in voltage.Image reconstruction for electrical impedance tomography(EIT)is an ill-posed and ill-conditioned problem.Therefore,the solutions for EIT are highly sensitive to the measurement noise.Some high-frequency information of reconstructed image is easy to be lost while image denoising.The sparse reconstruction method can effectively retain the key information and improve the quality of reconstructed images.At present,the global fixed dictionary is always adopted for sparse reconstruction algorithms of EIT.That means the feature extraction and sparse representation is performed for the whole image.The quality of reconstructed image based on sparse reconstruction method still needs to be improved.On the one hand,local detail information is ignored in the process of sparse representation;on the other hand,the mismatch between the dictionary atom and the structural features of the image may also cause the image representation to be less sparse.In order to solve above problems,a novel image reconstruction algorithm for EIT based on patch-based sparse representation is developed in this paper.The main work is as follows:(1)The block sparse method is introduced to extract the detail feature,which is ignored in global sparsity method.The reconstructed image is decomposed into multiple overlapping image blocks with the zero-fill method.Each image block is sparsely represented.As a result,the detailed structural features of the image can be accurately described.(2)An adaptive dictionary learning method for EIT sparse reconstruction is proposed.The Kernel Singular Value Decomposition(KSVD)algorithm is used to train the sparse dictionary of image block.The dictionary learning and image reconstruction are alternately performed.The reconstruction result of the new algorithm is evaluated through simulation experiments.Compared with the reconstructed results of traditional algorithms,the block sparse method can effectively remove the noise and artifacts of the reconstructed image and improve the quality of reconstructed image.(3)A dynamic experimental system is constructed to validate the proposed algorithm.EIT visualization software system based on Java is developed.With the aid of multi-threading technology,the parallel computing method based on thread pool is proposed.The solution process of the block sparse algorithm is optimized based on tread pool method,so that the process of image reconstruction is accelerated.In order to further improve the contrast of the impedance image,a color re-mapping method is proposed to enhance the visual effect of the reconstructed image.