A real-time correction and stitching algorithm for underwater fisheye images

• Published in: Signal, image and video processing Vol. 16; no. 7; pp. 1783 - 1791
• Main Authors: Wang, Zhanhua, Tang, Zhijie, Huang, Jingke, Li, Jianda
• Format: Journal Article
• Language: English
• Published: London Springer London 2022; Springer Nature B.V
• Subjects: Algorithms; Computer Imaging; Computer Science; Error correction; Fisheye views; Image Processing and Computer Vision; Image quality; Marine resources; Multimedia Information Systems; Original Paper; Pattern Recognition and Graphics; Screening; Signal,Image and Speech Processing; Stitching; Underwater vehicles; Vision; FAST-MIS stitching algorithm; Image stitch; Fisheye correction algorithm; Accurate calibration method
• ISSN: 1863-1703; 1863-1711
• DOI: 10.1007/s11760-022-02135-0

The application of underwater vehicles is an important link in the process of detecting marine resources, in the process, it is necessary to correct underwater images and stitch corrected images to receive a panoramic image. To implement this process, this paper proposes a fisheye image stitching algorithm include an accurate correction algorithm and a FAST-MIS stitching method to correction and stitch the fisheye image, finally obtain the stitch panoramic image. First of all, for correcting the fisheye image, this paper presents a accurate correction algorithm, which produces fewer errors and results in better image quality. In this paper, we use FAST-MIS (Fast multi-image stitching) stitching algorithm to quickly stitch multiple photographs. The FAST-MIS stitching algorithm. The algorithm proposed in this paper transforms the feature points in the process of large-size image Mosaic into small-size image feature points, which reduces the screening range of feature band, thus speeding up the screening speed and accuracy of effective feature points. Through experiment results shows that the error of the accurate correction algorithm proposed in this paper is reduced compared with the traditional method, and the correction effect of the result is also improved. Compared with the traditional ORB (Oriented FAST and Rotated BRIEF) method, the FAST-MIS stitching algorithm proposed in this paper can shorten the time of image stitch by about 23%.