Open AccessCIRCULARITY: A Metric for Region Template Selection in Airborne Vision Navigation
Author(s) -
Hua Hu,
Zelin Shi,
Yunpeng Liu
Publication year - 2019
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 - 10.1088/1742-6596/1325/1/012083
Subject(s) - metric (unit) , artificial intelligence , computer science , matching (statistics) , selection (genetic algorithm) , similarity (geometry) , enhanced data rates for gsm evolution , pattern recognition (psychology) , template matching , computer vision , ideal (ethics) , feature (linguistics) , algorithm , mathematics , image (mathematics) , statistics , engineering , philosophy , operations management , linguistics , epistemology
Reliability template is a key factor of airborne vision navigation system. This paper proposes a new metric for region template selection. Firstly, a measurement model for uncertainty and inaccuracy of edge feature is given in detail. Furthermore, the influence of spatial distribution of edge features of correct recognition probability is analyzed in statistical probability theory. Then an “ideal” spatial distribution of region template which is the most easy to be recognized is proved by extreme-case analysis. Finally, a new metric is proposed by calculating the shape distance from this “ideal” spatial distribution, called CIRCULARITY, which can be deemed as “geometry self-similarity” to one target. The validity of new metric is experimentally demonstrated by using real ground targets. The results show that the new metric can meet the requirements of matching area selection and generating template.