Work

A Monocular Vision-Based Method for Aeroengine Blade Damage Measurement via Keypoint Detection

Year: 2026

Type: article

Abstract: Aeroengine blade damage size is a critical factor in determining engine maintenance strategies, yet existing manual measurement methods face limitations in both efficiency and precision. Addressing this, this study proposes an automated vision-based measurement system for high-precision in situ quantification of blade damage under endoscopic conditions. The key contribution lies in a unified measurement framework that integrates learning-based perception with geometric modeling. Specifically, a deep learning module is designed to serve as a virtual sensing component, enabling robust extraction of measurement primitives, including damage boundaries and keypoints, from challenging borescope images. To tackle challenges posed by sparse damage targets, small scales, and strong background interference in borescope images, the detection transformer (DETR) framework is enhanced with a dynamic density estimation mechanism, incorporating multiexpert fusion, adaptive density guidance, and an intersection-over-union (IoU)-based query selection strategy, thereby significantly improving detection reliability. Meanwhile, a high-precision motion control device is designed to adjust the borescope pose. By coupling motion information with depth variation and introducing physical-size constraints, a stable mapping from pixel coordinates to real-world dimensions is established, enabling accurate estimation of damage geometry parameters. The proposed system forms a complete measurement pipeline from visual perception to physical quantification, achieving automatic identification and real-time measurement of blade damage. This work provides a practical solution for intelligent industrial inspection and demonstrates the potential of integrating artificial intelligence with measurement systems.

Source: IEEE Transactions on Instrumentation and Measurement

DOI: 10.1109/tim.2026.3716477

Authors: Huadong Shi, Pengtao Li, Hongfu Zuo, Siyan Cao, Huajuan Shi

Institutions: Henan University of Science and Technology, Nanjing University of Aeronautics and Astronautics, Zhengzhou University

Language: English

Topic: Industrial Vision Systems and Defect Detection

Subfield: Industrial and Manufacturing Engineering

Field: Engineering

Domain: Physical Sciences

Open Access status: hybrid

Estimated APC paid: $2,645

Funders: National Natural Science Foundation of China, Henan Provincial Science and Technology Research Project