基于计算机视觉的裂纹自动识别算法在飞机全尺寸疲劳试验中具有较好的工程应用前景。但由于飞机结构构型多样、疲劳试验环境复杂，直接应用现有的目标检测算法会存在较高的误判率。因此，提出一种基于关键部位状态对比的裂纹识别方法，以人脸识别模型FaceNet为基础，利用对比机制消除结构表面纹理、划痕等干扰因素的影响，并通过对裂纹数据结构和特征分布规律的分析，对FaceNet模型的样本生成规则、网络架构和损失函数进行了适应性改进。该方法具有对裂纹敏感、对图像质量要求低的特点。在疲劳试验环境中，该方法对长度为0.2~5 mm裂纹的检测准确率为97.6%，相较于现有方法优势明显。
The automatic crack identification algorithm based on computer vision has a good engineering application prospect in aircraft full-scale fatigue test. However, due to the diversity of aircraft structures and the complexity of fatigue test environment, the direct application of the existing target detection algorithm will have a high misjudgment rate. Therefore, this paper proposes a crack identification method based on state comparison of key structure. Based on the face recognition model FaceNet, contrast mechanism is used to eliminate interference of structure surface texture and scratches, and through the analysis of crack data structure and characteristic distribution law, sample generation rules, network architecture and the loss function of FaceNet are improved adaptively. The model is sensitive to cracks and has low demand on image quality. In the test environment, the detection accuracy of the proposed method is 97.6% for the crack length of 0.2-5 mm, which has obvious advantages over the existing methods.
[1] MCCANN D M, FORDE M C. Review of NDT methods in the assessment of concrete and masonry structures[J]. NDT & E International, 2001, 34(2):71-84.
[2] Nayler J L. Dictionary of aeronautical engineering[M]. London:Newnes, 1959:25-29.
[3] VAVILOV V P, DERUSOVA D, CHULKOV A, et al. Inspecting aviation composites at the stage of airplane manufacturing by applying ‘classical’ active thermal NDT, ultrasonic IR hermography and laser vibrometry[C]//Thermosense:Thermal Infrared Applications XL. Orlando, FL:SPIE, 2018:106-112.
[4] KOMSKY I N, ACHENBACH J D, HAGEMAIER D, et al. A computerized self-compensating system for ultrasonic inspection of airplane structures[J]. NDT:A Partner in Engineering Innovation,1994, 27(3):77-85.
[5] REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[6] REDMON J, DIVVALA S, GIRSHICK R, et al. You Only Look Once:unified, real-time object detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2016:779-788.
[7] DENG J H, LU Y, LEE V C S. Imaging-based crack detection on concrete surfaces using You Only Look Once network[J]. Structural Health Monitoring, 2021, 20(2):484-499.
[8] DU Y C, PAN N, XU Z H, et al. Pavement distress detection and classification based on YOLO network[J]. International Journal of Pavement Engineering, 2021, 22(13):1659-1672.
[9] WANG S Y, ZHANG P Z, ZHOU S Y, et al. A computer vision based machine learning approach for fatigue crack initiation sites recognition[J]. Computational Materials Science, 2020, 171:109259.
[10] 杨晶晶, 李鸿宇, 王子睿, 等. 基于单步目标识别架构的轻量级裂纹图像自动识别算法[J]. 固体火箭技术, 2020, 43(5):648-653. YANG J J, LI H Y, WANG Z R, et al. Lightweight crack image automatic recognition algorithm based on single-stage object detecting[J]. Journal of Solid Rocket Technology, 2020, 43(5):648-653(in Chinese).
[11] HENRIQUES J F, CASEIRO R, MARTINS P, et al. High-speed tracking with kernelized correlation filters[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(3):583-596.
[12] DANELLJAN M, KHAN F S, FELSBERG M, et al. Adaptive color attributes for real-time visual tracking[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2014:1090-1097.
[13] LI Y, ZHU J K. A scale adaptive kernel correlation filter tracker with feature integration[C]//Computer Vision-ECCV 2014 Workshops.[S.l.]:Springer Cham, 2014:254-265.
[14] TU Z W, CHEN X R, YUILLE A L, et al. Image parsing:unifying segmentation, detection, and recognition[J]. International Journal of Computer Vision, 2005, 63(2):113-140.
[15] ZHANG Z P, PENG H W. Deeper and wider siamese networks for real-time visual tracking[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2019:4586-4595.
[16] 龚轩,乐孜纯,王慧, 等.多目标跟踪中的数据关联技术综述[J].计算机科学,2020,47(10):136-144. GONG X, LE Z C, WANG H, et al. Overview of data association techniques in multi-target tracking[J]. Computer Science, 2020, 47(10):136-144(in Chinese).
[17] BOSE B, WANG X G, GRIMSON E. Multi-class object tracking algorithm that handles fragmentation and grouping[C]//2007 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2007:1-8.
[18] SCHROFF F, KALENICHENKO D, PHILBIN J. FaceNet:A unified embedding for face recognition and clustering[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2015:815-823.
CJA