The regularization parameters in image reconstruction of oil-air lubrication ECT system were determined by U-curve method,the ill-conditioning of sensitivity matrix after regularization was analyzed. The first image reconstruction of the pipeline section of oil-air lubrication ECT system was completed by LBP algorithm and Tikhonov regularization algorithm respectively. The secondary image reconstruction of the pipeline section was completed after the threshold filtering threshold was optimized for gray distribution matrix of the first reconstructed image. The results show that the regularization parameters selected by U-curve method have a significant effect on weakening the ill-conditioning of sensitivity matrix compared with L-curve method. In the first image reconstruction,the image reconstruction quality is greatly improved,and in the secondary image reconstruction after threshold filtering threshold optimization,the image reconstruction quality is further improved. The research results show that both of the regularization parameters determined by U-curve method and threshold filter threshold optimization can improve the image reconstruction quality of oil-air lubrication ECT system.