An adaptive image-stitching algorithm for an underwater observation system

Published: 08th May 2020
Views: N/A

Currently the majority of countries are devoted to focusing on underwater tracking. However, the acquisition of underwater images is performed in noisy surroundings with low visibility because natural light is unavailable and if man-made light is used, the observable range is limited [1]. Acoustic cameras can provide extremely high resolutions (for sonar) and accelerated refresh rates [2]. Therefore, sonar systems are broadly applied to get images of the seabed or another underwater objects. It's not conducive to tracking objects or environments submerged through sonar gear alone since the observation angle of high resolution sonar gear is so little that only a part of the submerged scene can be observed, i.e., the horizontal view angle of DIDSON (dual-frequency identification sonar) is 28.8[degrees]. Hence, most sonar equipment is often installed in rotational mechanism to have distinct scene advice; as such, sonar images are stitched to enlarge the observation horizon of the underwater environment [3]. Therefore, image mosaicing technology plays an important role in observation systems that are submerged. There has been much research into optical image-stitching; however, little research has been given to stitching for sonar pictures. Image stitching on an optical picture could be split into algorithms based on the pixel-level, the frequency power spectrum and features [4]. The calculation of the pixel-level algorithm is simple and quick, but the mosaicing result is not ideal if there aren't enough features or when the pictures are particularly noisy. The computation of the frequency power spectrum algorithm is also not slow and noise correlation interference may be overcome, but it needs a sufficient overlap width. Meanwhile, attribute points can certainly be pulled by the feature algorithm, which is suitable for stitching the images with enough features but the cost of a heavy calculation load. It is inappropriate to adopt a pixel-level algorithm for stitching sonar images with plenty of sound. In the view of present study, the phase correlation method [5-7] is the most popular algorithm in the field of the frequency power spectrum algorithm while the SIFT (scale invariant feature transform) [8], FAST (attributes from gifted section evaluation) [9] and ORB (oriented RAPID and rotated BRIEF) [10] and SURF [11-13] algorithms have great popularity in the area of feature algorithms. The algorithms enable the measurement of rotation translation and scaling variables between two pictures. The stitching is not bad on paper. However, the precision of the stitching result for sonar images with adequate attributes is lower than for the attribute algorithm. The SIFT algorithm continues to be analyzed for pulling invariant features that were identifying from pictures which may be used to do trusted matching between different views of an object or scene [8]. Because the SIFT feature points are described as 128 -dimensional vectors, the speed of the algorithm isn't fast. Mair et al. [9] have described a novel corner detection strategy, called the FAST (attributes from accelerated section test) algorithm. The FAST algorithm only uses the surrounding pixels advice to get the attribute points, which can be easy and relatively fast. But, the truth of the stitching result from this algorithm is not better than for the SIFT algorithm.

Report this article Ask About This Article

More to Explore