Graded fuzzy edge detection for imperceptibility optimization of image steganography

Abstract

ABSTRACT The edge area of an image is more resilient to distortion caused by embedding in steganography, driving the advancement of edge detection-based methods. State-of-the-art techniques, including hybrid, dilation, and Fuzzy-based approaches, have been developed to enhance steganography performance. Typically, these methods categorize image pixels into two regions: edge and non-edge areas. This research introduces graded Fuzzy edge detection, classifying pixels into main, supporting, and non-edge categories. Consequently, the embedding priority and message bit determination become more diverse and adaptive. Experimental results demonstrate that graded Fuzzy edge detection optimizes imperceptibility while preserving security and capacity. This technique is combined with various steganographic and reversible data hiding (RDH) methods, exhibiting improvements in stego image quality with over 1dB enhancement based on PSNR, as well as maintaining payload capacity and security aspects according to RS and PDH steganalysis.