Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Autoencoded Image Compression for Secure and Fast Transmission

Aryan Kashyap Naveen, Sunil Thunga, Anuhya Murki, Mahati A Kalale, Shriya Anil

TL;DR

The paper addresses secure and efficient image transmission by using a convolutional autoencoder for end-to-end compression. It introduces a composite loss that combines reconstruction accuracy with a residual term to preserve high-frequency details, achieving a 12:1 compression ratio while maintaining high visual quality (SSIM ≈ 0.975). The latent-space representation inherently encrypts the data, as decoding requires the trained decoder at the destination, and a progressive training scheme with residual outputs reduces training time. Empirical results show substantial latency reductions (~87.5%) and competitive image quality compared with traditional codecs, with potential for further gains through quantisation and entropy coding.

Abstract

With exponential growth in the use of digital image data, the need for efficient transmission methods has become imperative. Traditional image compression techniques often sacrifice image fidelity for reduced file sizes, challenging maintaining quality and efficiency. They also compromise security, leaving images vulnerable to threats such as man-in-the-middle attacks. This paper proposes an autoencoder architecture for image compression to not only help in dimensionality reduction but also inherently encrypt the images. The paper also introduces a composite loss function that combines reconstruction loss and residual loss for improved performance. The autoencoder architecture is designed to achieve optimal dimensionality reduction and regeneration accuracy while safeguarding the compressed data during transmission or storage. Images regenerated by the autoencoder are evaluated against three key metrics: reconstruction quality, compression ratio, and one-way delay during image transfer. The experiments reveal that the proposed architecture achieves an SSIM of 97.5% over the regenerated images and an average latency reduction of 87.5%, indicating its effectiveness as a secure and efficient solution for compressed image transfer.

Autoencoded Image Compression for Secure and Fast Transmission

TL;DR

The paper addresses secure and efficient image transmission by using a convolutional autoencoder for end-to-end compression. It introduces a composite loss that combines reconstruction accuracy with a residual term to preserve high-frequency details, achieving a 12:1 compression ratio while maintaining high visual quality (SSIM ≈ 0.975). The latent-space representation inherently encrypts the data, as decoding requires the trained decoder at the destination, and a progressive training scheme with residual outputs reduces training time. Empirical results show substantial latency reductions (~87.5%) and competitive image quality compared with traditional codecs, with potential for further gains through quantisation and entropy coding.

Abstract

With exponential growth in the use of digital image data, the need for efficient transmission methods has become imperative. Traditional image compression techniques often sacrifice image fidelity for reduced file sizes, challenging maintaining quality and efficiency. They also compromise security, leaving images vulnerable to threats such as man-in-the-middle attacks. This paper proposes an autoencoder architecture for image compression to not only help in dimensionality reduction but also inherently encrypt the images. The paper also introduces a composite loss function that combines reconstruction loss and residual loss for improved performance. The autoencoder architecture is designed to achieve optimal dimensionality reduction and regeneration accuracy while safeguarding the compressed data during transmission or storage. Images regenerated by the autoencoder are evaluated against three key metrics: reconstruction quality, compression ratio, and one-way delay during image transfer. The experiments reveal that the proposed architecture achieves an SSIM of 97.5% over the regenerated images and an average latency reduction of 87.5%, indicating its effectiveness as a secure and efficient solution for compressed image transfer.
Paper Structure (18 sections, 3 equations, 8 figures, 2 tables)

This paper contains 18 sections, 3 equations, 8 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Literature review
  3. Dataset
  4. Training Dataset
  5. Evaluation Dataset
  6. Architecture
  7. Encoder
  8. Decoder
  9. Training
  10. Model and Setup
  11. Hyperparameters
  12. Loss Calculation
  13. Results
  14. Effect of the Residual Component
  15. Performance Evaluation Across Varied Batch Sizes using Progressive Image Compression (PgIC)
  16. ...and 3 more sections

Figures (8)

  • Figure 1: Autoencoder Architecture
  • Figure 2: Model setup during training
  • Figure 3: Model setup during testing
  • Figure 4: Loss vs Epochs during training
  • Figure 5: Comparison of results with and without Progressive Image Compression
  • ...and 3 more figures