Premium
The algorithm of seamless image mosaic based on A‐KAZE features extraction and reducing the inclination of image
Author(s) -
Qu Zhong,
Bu Wei,
Liu Ling
Publication year - 2018
Publication title -
ieej transactions on electrical and electronic engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.254
H-Index - 30
eISSN - 1931-4981
pISSN - 1931-4973
DOI - 10.1002/tee.22507
Subject(s) - scale invariant feature transform , image stitching , artificial intelligence , algorithm , computer vision , computer science , gaussian filter , feature detection (computer vision) , robustness (evolution) , image fusion , feature extraction , pattern recognition (psychology) , image processing , image (mathematics) , biochemistry , chemistry , gene
The traditional feature point detection algorithm is based on the linear scale decomposition. In the SIFT (Scale Invariant Feature Transform) algorithm, features are obtained through building the image pyramid by the Gaussian filter. SIFT has good robustness but has some flaws as well. Gaussian filter neither preserve object boundaries nor smooth the same level details and noise at all scales, which impair the accuracy and distinctiveness of the feature point positioning. Nonlinear scale decomposition can solve these problems. In this paper, a new image mosaic algorithm based on A‐KAZE feature is proposed to take advantages of the A‐KAZE algorithm in terms of rotation invariance, illumination invariance, speed, and stability. The optimal stitching line is obtained and the multi‐resolution fusion algorithm is used to fuse the image in order to achieve a satisfactory seamless image of high resolution. The whole straightening method is applied in the image mosaic to solve the problem of the tilt of multiple image mosaic. Experimental results show that stitching algorithm in this paper is faster and more robust compared to the traditional SIFT algorithm. © 2017 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.