Open AccessBlur-robust image registration and stitching
Author(s) -
Yann Do,
Rustam Paringer,
Alexander Kupriyanov,
Yegor Goshin
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 - 10.1088/1742-6596/1368/5/052043
Subject(s) - image stitching , computer science , artificial intelligence , computer vision , key (lock) , process (computing) , matching (statistics) , task (project management) , image registration , point (geometry) , image (mathematics) , mathematics , engineering , statistics , geometry , computer security , operating system , systems engineering
This paper addresses the functioning of a new key point detection method for images. Such methods are useful in applications as image reconstructions, object signature recognition or video stabilization. The method, Blurred Image Matching (BIM), is based on the comparison of large areas of interest in images. BIM uses a process of blurring and shape comparison that makes the approach unique and in the vast majority of tested cases results in higher performances than other approaches. Among BIM’s key features are the amount of points used for comparison, 19 to 129 times fewer than other methods and of higher quality, allowing faster processing. Another of its key features is a high efficiency in the comparison of images, with a success rate higher than the average of existing methods by 13% to 36%, depending on the images’ characteristics. The method is still in a development stage and will significantly improve, however the results obtained so far are extremely promising, its task solving approach is new and makes a new range of applications necessiting key point detection possible. All this allows us to foresee BIM as a future standard among such methods.