Open AccessResearch on Algorithm of fighter landing gear in bad video image
Author(s) -
Manshu Tang,
Wei Fang,
Wenjun Yan,
Qing Ling
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2024/1/012008
Subject(s) - artificial intelligence , computer science , computer vision , image (mathematics) , cluster analysis , feature (linguistics) , process (computing) , dimension (graph theory) , feature extraction , mathematics , philosophy , linguistics , pure mathematics , operating system
The retracted and retracted state of landing gear is a problem that needs to be paid close attention to in the landing process of fighter. The traditional visual observation is easy to be affected by the climate environment. Aiming at the problems of small target, low definition and difficult detection of fighter landing gear in bad video images, a YOLOv4 target detection algorithm based on image enhancement is proposed to detect fighter landing gear. In order to improve the accuracy of landing gear detection, ACE algorithm is first applied to the video image to enhance the video image, then the main feature extraction network of YOLOv4 is used to extract the feature information, and K-means clustering algorithm is used to cluster and analyze the number and aspect ratio dimension of target candidate frames. Experiments on the dataset show that the detection accuracy after image enhancement reaches 88.12%, which is 1.19% higher than the original image; The recall rate reached 73.44%, an increase of 4.17%; The value of mAP reached 82.65%, an increase of 1.08%.