Aircraft Skin Damage Detection and Assessment From UAV Images Using GLCM and Cloud Model

The traditional manual detection of aircraft skin is subjective and inefficient. In order to achieve rapid damage detection, this paper proposes an aircraft skin damage detection and evaluation framework by combining gray level co-occurrence matrix (GLCM) and cloud model. In the experimental stage, the UAV picks up the damage image of the aircraft, and the texture feature data is output by using the GLCM algorithm. The introduction of the cloud model evaluation system makes the skin damage type be quickly judged. The results show that the proposed method has good recognition ability for aircraft skin damage, and the identification accuracy of the verification image set reaches 85%. In the validation image set, 50 % of the corrosion spalling images were judged to be normal, which may be due to the similarity of the two types of texture features, and also indicates that the initial stage of skin damage starts from pitting. When evaluating the damage image, 59 % of the cloud droplets fall in the normal level, indicating that the damage is not serious, and the damage maintenance of the aircraft can be delayed according to the usage.