Enhanced edge-based steganography using image segmentation and random LSB substitution for secure data hiding

Steganography offers completely invisible and secure communication that relies on careful embedding so that it is impossible for even keen observers to detect hidden data. In this work, we propose an enhanced edge-based framework that brings together image segmentation, Canny edge detection, and a 128-bit seeded pseudo-random number generator (PRNG) to drive randomized least significant bit (LSB) substitution. Edge detection plays a critical role in identifying regions of significant intensity variation in digital images, which can be leveraged to enhance the imperceptibility and robustness of steganographic techniques. This paper proposes an improved edge-based steganographic method that integrates image segmentation with randomized LSB substitution to embed confidential data securely. The approach begins by segmenting the input image and extracting edge regions using the Canny edge detector. These edge pixels are then randomized using a 128-bit pseudo-random key to determine the embedding locations, ensuring unpredictability and increased security. To maintain high visual fidelity, the secret message is divided into uniform-length binary segments and embedded only within the selected edge pixels, thereby minimizing distortion. The proposed method is evaluated using key performance metrics, including peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), embedding capacity, and computational efficiency. Experimental results on standard benchmark images demonstrate high imperceptibility (average PSNR exceeding 65 dB and SSIM approaching 1.00), low embedding time, and strong resistance to statistical detection. Histogram analysis and ablation studies further confirm the visual and statistical invisibility of the stego-images. Overall, the method offers a robust, secure, and efficient solution for data hiding in digital images.