EUVIMEDEUVIMED entdecken
Crossref

Underwater Image Transmission Model Using Improvised Feed Forward Fully Connected Deep Neural Network

Sumanth Sakkara, Cyril Prasanna Raj P.

Journal of Innovative Image Processing · 21.07.2026 · 8 (3) · 1144-1169

DOI: 10.36548/jiip.2026.3.019 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Innovative Image Processing, online erschienen am 21.07.2026. Verfasst von Sumanth Sakkara und 1 weiteren Autor:innen. Die Arbeit nennt 37 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Underwater security surveillance in coastal regions of India is of paramount importance. Underwater drones are deployed and enabled with acoustic sensors and cameras to detect and identify foreign objects. However, reliable underwater image transmission remains challenging due to channel noise and limited underwater acoustic bandwidth. In this paper, a novel method for the transmission of images captured underwater to base stations with minimum loss and robust against underwater noise is developed. The proposed Deep Neural Network (DNN) model, comprising multiple stages of neural network structure, processes the input image and compresses it into a data stream generated using QAM schemes to be transmitted over an underwater channel. The DNN structure at the receiver reconstructs the data stream into a visible image with minimal noise. The developed model was evaluated for performance considering different underwater channel conditions with SNR settings from -10 dB to + 10 dB. The BER metric was estimated to vary between 10⁻² to 3.4×10⁻³, with an SSI measurement of 0.9235 and entropy measuring 93.44%. The reconstructed images were analyzed with the PSNR metric, and the proposed model demonstrates an improvement of 51.48%. The proposed framework achieves up to 93.75% compression, reducing bandwidth requirements for underwater acoustic communication while preserving image quality. The simulation results were obtained using the MATLAB environment by modeling the proposed DNN model and training was carried out considering 3590 images. The end-to-end model was evaluated for its performance and effectiveness, demonstrating its suitability for reliable image transmission through underwater channels.

Zitieren

APA

Sakkara, S., & P., C. P. R. (2026). Underwater Image Transmission Model Using Improvised Feed Forward Fully Connected Deep Neural Network. Journal of Innovative Image Processing, 8(3), 1144-1169. https://doi.org/10.36548/jiip.2026.3.019

Vancouver

Sakkara S, P. CPR. Underwater Image Transmission Model Using Improvised Feed Forward Fully Connected Deep Neural Network. Journal of Innovative Image Processing. 2026;8(3):1144-1169. doi:10.36548/jiip.2026.3.019

BibTeX
@article{Sakkara2026,
  author = {Sakkara, Sumanth and P., Cyril Prasanna Raj},
  title = {Underwater Image Transmission Model Using Improvised Feed Forward Fully Connected Deep Neural Network},
  journal = {Journal of Innovative Image Processing},
  year = {2026},
  volume = {8},
  number = {3},
  pages = {1144--1169},
  doi = {10.36548/jiip.2026.3.019}
}
Kostenlos starten →