Aiming at the problem of accurately detecting and quantifying scratches and crater defects in compressor blades with existing methods, an algorithm based on structured light point cloud data is proposed. First, an IDW-NA point cloud feature enhancement algorithm, which integrates inverse distance weighted curvature and normal angle of large and small regions, is used to highlight the defects. In the defect localization process, the Otsu method (OTSU) is innovatively introduced to eliminate the limitations of manually setting thresholds, followed by the Z-score-based defect integrity expansion (ZDE) algorithm to achieve complete segmentation of the defects. Finally, the PCA algorithm is improved to perform quantitative analysis of the defects. Experimental results show that, compared to existing algorithms, the proposed method provides better performance in terms of defect segmentation integrity and continuity. The average absolute error of the final segmented defect size is no more than 0.105 mm, and the average percentage error is no more than 7.27%, confirming the accuracy and effectiveness of this approach.