Hybrid Deep Learning-Based Edge Detection for Secure Image Steganography

Steganography is a technique used to conceal secret information in digital assets such as images. One common method is to embed data along the edges of an image, where changes are harder to notice. This strategy helps maintain the image's natural appearance while protecting the hidden information. Traditional edge detection methods like Canny, Sobel, and Laplacian of Gaussian (LoG) generally detects extra or incorrect edges, especially into smooth or very detailed areas. As a result, the quality of data embedding may be compromised. It leads to reduced image quality. To solve this problem, researchers have now started using deep learning in steganography. Deep learning models work better at understanding image features and are able to find more accurate edges. This paper presents a hybrid deep learningbased edge detection method for steganography. It combines three deep learning-based edge detectors using a bitwise AND operation. These three edge detectors are Holistically-Nested Edge Detection (HED), Dense Extreme Inception Network for Edge Detection (DexiNed) - using both average and fused outputs —, and Pixel Difference Network (PiDiNet). This approach helps to find the most accurate edges and avoids hiding data in the wrong places. The performance of the proposed method is evaluated on a set of three standard test images with diverse textures. A comprehensive analysis is conducted under different payload capacities to assess the robustness of the technique. Experimental results shows that our method keeps the image quality high and gives better Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) values compared to traditional methods. It allows a large amount of data to be securely embedded. Overall, the proposed method provides a reliable deep learning solution for secure information hiding and protection of digital content.

• Publication year

  2025

• Venue

  2025 Second International Conference on Pioneering Developments in Computer Science & Digital Technologies (IC2SDT)

• Publication date

  2025-12-04

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1109/IC2SDT68218.2025.11383731
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Image Semantic Steganography: A Way to Hide Information in Semantic Communication

  2025cited by this paper
• Payload-Independent Direct Cost Learning for Image Steganography

  2024cited by this paper
• An Effect of Huffman Encoding with Hybrid Edge Detection in LSB Image Steganography

  2024cited by this paper
• A secure image steganography based on LSB technique and 2D chaotic maps

  2024cited by this paper
• A Novel Dual Image Reversible Data Hiding Scheme Based on Vector Coordinate With Triangular Order Coding

  2024cited by this paper
• Graded fuzzy edge detection for imperceptibility optimization of image steganography

  2023cited by this paper
• Image steganography based on difference of Gaussians edge detection

  2023cited by this paper
• A Novel Hybrid Edge Detection and LBP Code-Based Robust Image Steganography Method

  2023cited by this paper
• Recent advances on image edge detection: A comprehensive review

  2022cited by this paper
• A Novel Steganography Technique for Digital Images Using the Least Significant Bit Substitution Method

  2022cited by this paper
• Adaptive Image Steganography Using Fuzzy Enhancement and Grey Wolf Optimizer

  2022cited by this paper
• A secure data hiding approach based on least-significant-bit and nature-inspired optimization techniques

  2022cited by this paper
• Dense Extreme Inception Network for Edge Detection

  2021cited by this paper
• Pixel Difference Networks for Efficient Edge Detection

  2021cited by this paper
• A New Adversarial Embedding Method for Enhancing Image Steganography

  2021cited by this paper
• End-to-End Image Steganography Using Deep Convolutional Autoencoders

  2021cited by this paper
• Image steganography using deep learning based edge detection

  2021cited by this paper
• Hiding Images within Images

  2020cited by this paper
• Improved payload capacity in LSB image steganography uses dilated hybrid edge detection

  2019cited by this paper
• Reversible Data Hiding in Permutation-based Encrypted Images with Strong Privacy

  2019cited by this paper
• Edge Based Image Steganography: Problems and Solution

  2019cited by this paper
• High fidelity based reversible data hiding using modified LSB matching and pixel difference

  2019cited by this paper
• Edge based image steganography with variable threshold

  2018cited by this paper
• High payload image steganography based on Laplacian of Gaussian (LoG) edge detector

  2018cited by this paper
• Very High Capacity Image Steganography Technique Using Quotient Value Differencing and LSB Substitution

  2018cited by this paper
• StegNet: Mega Image Steganography Capacity with Deep Convolutional Network

  2018cited by this paper
• An improved secure image hiding technique using PN-sequence based on DCT-OTP

  2017cited by this paper
• A high payload steganographic algorithm based on edge detection

  2017cited by this paper
• Hiding Images in Plain Sight: Deep Steganography

  2017cited by this paper
• Holistically-Nested Edge Detection

  2015cited by this paper
• Analysis of LSB based image steganography techniques

  2001cited by this paper
• Secure blind image steganographic technique using discrete Fourier transformation

  2001cited by this paper
• Exploring steganography: Seeing the unseen

  1998cited by this paper
• An Isotropic 3×3 image gradient operator

  1990cited by this paper
• A Computational Approach to Edge Detection

  1986cited by this paper
• Theory of edge detection

  1979cited by this paper

• No citing papers are available for this paper.