Subsea pipelines are prone to developing cracks during long-term service, and traditional methods often suffer from low localization accuracy and poor stability in complex noisy environments. This study investigates defect localization based on an improved cross-correlation time delay estimation method for acoustic emission (AE) signals, which is significant for identifying and localizing sudden AE events under challenging conditions. AE detection technology, as a dynamic, real-time, and non-destructive evaluation method, captures transient elastic waves generated by internal structural changes due to stress or environmental variations, enabling early warning of potential damage and accurate health monitoring of structures. Experimental results demonstrate that the improved cross-correlation time delay method effectively enhances the localization accuracy for AE sources, achieving an average localization error of 0.0133 m and a relative error of 2.67%, which sufficiently meets practical requirements. The findings indicate that this method possesses high practical value for damage detection and localization analysis in complex structures.