An adaptive image-stitching algorithm for an underwater monitoring system

Published: 08th May 2020
Views: N/A

Now the majority of countries are devoted to focusing on submerged observation. However, the acquisition of underwater images is conducted with low visibility in noisy surroundings because natural light is unavailable and if man-made light is applied, the visible range is restricted [1]. Acoustic cameras can provide extremely high resolutions (for sonar) and accelerated refresh rates [2]. Therefore, sonar systems are widely employed to get images of the seabed or another underwater objects. It is not contributory to monitoring items or surroundings underwater through sonar gear alone because the observation angle of high resolution sonar gear is so small that only element of the underwater scene can be found, i.e., the horizontal view angle of DIDSON (dual-frequency identification sonar) is 28.8[degrees]. Hence, most sonar gear is frequently installed to obtain different scene info; as such, sonar pictures are stitched to enlarge the observation horizon of the submerged environment [3]. Thus, image mosaicing technology plays an important role in monitoring systems that are submerged. There's been much research into optical image-stitching; nevertheless, little studies have been devoted to stitching for sonar images. Image stitching on an optical image may be separated into algorithms based on the pixel-level, the frequency power spectrum and features [4]. The calculation of the pixel-level algorithm is quick and easy, but the result that is mosaicing is not perfect if the pictures are particularly noisy or if there aren't enough features. The calculation of the frequency power spectrum algorithm is also not slow and noise correlation interference may be beat, but it needs a sufficient overlap width. Meanwhile, attribute points can quickly be taken out by the feature algorithm, which is appropriate for stitching the images with enough features but the cost of a substantial calculation load. It is inappropriate to adopt a pixel-level algorithm for stitching sonar images with a great deal of noise. In previous studies, many researchers have concentrated on the frequency power spectrum algorithm or the characteristic algorithm independently to mosaic sonar picture. Several modified stage correlation algorithms on the basis of the Fourier transform have been proposed in the literature [57]. The algorithms enable the measurement of rotation translation and scaling factors between two images. The stitching resulting from this algorithm is good on paper. On the other hand, the precision of the stitching result for sonar pictures with features that are adequate is lower than for the attribute algorithm. The SIFT algorithm has been studied for pulling distinctive invariant attributes from images which may be used to perform reliable fitting between different views of an object or scene [8]. Since the SIFT feature points are described as 128 -dimensional vectors, the speed of this algorithm is slow. Mair et al. [9] have described a novel corner detection strategy, called the FAST (features from accelerated segment test) algorithm. The FAST algorithm only uses the surrounding pixels information to get the attribute points, which can be relatively fast and easy.

Report this article Ask About This Article

More to Explore