Underwater video mosaicing using topology and superpixel-based pairwise stitching

摘要: Abstract Expert and intelligent systems based on computer-vision algorithms are becoming a common part of our daily lives, as they help us solve problems in areas such medicine, agriculture, transportation, ecology. In this paper, we focus the application computer vision marine Here, propose new algorithm for mosaicing images coral reefs captured by scuba divers using hand-held cameras during reef surveys. Such often capture partial views surveyed area that then collated into mosaics assemblages. Accurate will coral-reef researchers rapidly assess not only status reefs, but also determine rates recovery or decline longitudinal data. Most standard distort images, however, moving objects parallax errors, which usually happen underwater hinder mosaic construction. To overcome these issues, uses two-step approach first detects removes loops low-quality regions from an estimated camera trajectory. The stitches remaining warping parallax-affected image with geometric transformation different one used parallax-unaffected regions. We tested method large spatial survey sites throughout Caribbean region. Our obtained better-quality than other state-of-the-art algorithms.