Abstract: Ultrasonic guided wave technology is widely used in structural nondestructive testing due to its long propagation distance and sensitivity to small damage. However, when there are multiple defects nearly in the pipeline, the reflected waves of the defects will overlap each other, deducing comparatively low localization resolution. In order to enhance the localization resolution, a sparse decomposition approach based on iterative optimization for pipeline detection is was firstly established. Then, under strong leakage noise interference, the overlapped wave packets are were separated. The reconstructed wave packets showed better temporal resolution, and subsequently improved the localization resolution. Compared with the conventional positioning method, the proposed algorithm verified by experiment can effectively identify two adjacent defects under strong leakage interference noise, and the positioning errors are 3. 33% and 4. 62% respectively.