Electromagnetic tomography (EMT) technology is used to realize the visualization of metal defects, which overcomes the lack of visualization of traditional testing technology. Firstly, a new planar sensor is designed. Secondly, according to the sparsity of defect distribution, the l1 regularized sparse imaging algorithm is proposed. The l1 regularization algorithm effectively overcomes the excessive smooth problem associated with traditional l2 regularization algorithm, whose imaging results are more accurate. Finally, in order to further prove the superiority of the new algorithm compared with l2 regularization algorithm, the simulation and experiment are conducted. The results show that sparse imaging algorithm can effectively improve the quality and accuracy of the defects images.