Abstract: Due to no requirements for wave velocity and time of flight of ultrasonic guided waves, probability-based diagnostic imaging (PDI) algorithm is especially suitable for damage identification of composite structures. However, the defect distribution probability of PDI algorithm is relatively inaccurate, which reduce the damage localization accuracy. Thus in engineering applications, the effectiveness of PDI algorithm could be affected. In order to improve the damage localization accuracy, an improved PDI algorithm was proposed. Using the relationship between the damage index and the relative distance from damage to the direct path of actuator-sensor path, the defect distribution probability of PDI algorithm was modified in the proposed algorithm. The validity of the proposed algorithm was assessed by identifying damages at different locations on a stiffened composite panel. The results show that the proposed algorithm can identify a single damage of stiffened composite panel accurately, and it can identify two damages effectively as well.