.

ISSN 2063-5346
For urgent queries please contact : +918130348310

Underwater Image Enhancement Using Deep Learning

Main Article Content

Mr.S.Balaji , J.Vengatesan , S.Satheesh kumar , K.Eagambaram
» doi: 10.31838/ecb/2023.12.si6.401

Abstract

The study of searching the ocean floor for both living things and non-living things is known as underwater archaeology. These kinds of studies are only achievable using sonar technologies. Sound is used as a source by the SONAR (SOund Navigation and Ranging) technology to create acoustic images of objects under the surface. Sonar comes in a variety of forms, and side scan sonar is one of the most accessible methods for visualising the bottom. The side scan sonar photos that are taken are frequently noisy, low contrast, and of poor visual quality. The acoustic images are displayed as grey scale images with different light and dark contrasts. This project's objective is to develop original picture-enhancing algorithms for object recognition in acoustic underwater images. The algorithms for image improvement include edge-preserving interpolation, intensity enhancement, contrast enhancement, and noise removal or denoising for object detection in audio pictures. Edgetech 4125 side scan sonar was employed in this research project to obtain the sound images. These images are affected by the speckle noise of the device. "SONAR" is an acronym for "Sound Navigation and Ranging." Sonar uses echoes to find and locate items beneath the sea, much to how porpoises and other marine animals do with their natural sonar systems. The suggested denoising method used Stationary Wavelet Transform and its shrinkage function to eliminate the speckle noises from the photos. Image enhancing methods were applied to the denoised photos. Two approaches based on multiresolution techniques were suggested. The Gaussian and Laplacian pyramid is used for intensity enhancement in the first enhancement technique. The second technique enhances the contrast of the photos using a Stationary Wavelet Transform and filtering methods. The suggested edge-based vi interpolation technique uses the improved image as input to increase the spatial resolution or size of the output images. Edge-based segmentation and object tracing algorithms are developed for the purpose of detecting objects in underwater sound pictures.

Article Details