A Method of Identifying Pitot Tube Cover Based on Color Histogram Features and SVM

The pitot tube cover is a ground protection device used to avoid the error of airspeed indication caused by foreign objects entering the pitot tube during the parking period and thus threatening the aviation safety. At the same time, the aircraft taking off without taking off the pitot tube cover is also a serious threat to aviation safety, so a method for identifying this kind of production defects is proposed. The existing 36 pitot tube on-site images (with and without covers) were preprocessed using the improved sliding window object detection algorithm and the image enhancement method based on the grayscale transformation. The color histogram features of the processed images was extracted. Using SVM for fault recognition, and the parameters of SVM are optimized by K-CV method. After optimization, the cross validated identification rate can up to 96%. The proposed method can be applied to the inspection of UAV around the aircraft, which is conducive to improving the PHM capability of civil aircraft pitot-static pressure system.