Open AccessInfrared small target detection based on robust principal component analysis joint directional derivative penalty
Author(s) -
Tiancheng Zhang,
Yiquan Wu,
Fei Zhou
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 - 10.1088/1742-6596/1873/1/012005
Subject(s) - principal component analysis , robust principal component analysis , robustness (evolution) , pattern recognition (psychology) , clutter , computer science , artificial intelligence , directional derivative , minification , norm (philosophy) , matrix norm , feature (linguistics) , computer vision , mathematics , algorithm , mathematical optimization , radar , eigenvalues and eigenvectors , telecommunications , mathematical analysis , biochemistry , chemistry , physics , linguistics , philosophy , quantum mechanics , political science , law , gene
Many robust principal component analysis based methods have been proposed for infrared small target detection recently. However, due to the lack of local prior information, these methods show poor performance under complex backgrounds. In this paper, a novel infrared patch-image model joint the directional derivative saliency feature is proposed. First, Laplace norm minimization is used instead of nuclear norm minimization to approximate low rank background can automatically capture the inherent rank information, which has higher matching with the real rank. Then, considering that the l 1 -norm constrains target patch-image inexactly, we construct a saliency map by using the directional derivative feature of the target and incorporate it into the l 1 -norm to enhance target and suppress background clutter simultaneously. Extensive experiments show the superiority of the proposed method to the other state-of-the-art methods in robustness and efficiency.