An unsolvable pixel reduced authentication method for color images with grayscale invariance

Abstract

This paper introduces an advanced grayscale-invariant authentication method for color images. Prior work directly utilizes the most significant bits (MSBs) of the red and blue channels for authentication code generation and embeds these codes into the least significant bits (LSBs). Concurrently, the green channel is adjusted to sustain consistent grayscale values. However, this direct utilization of MSBs often leads to a decline in image quality and results in a significant number of unsolvable pixels. To address these challenges, we present a new approach specifically designed to combat the deterioration of image quality and the prevalence of unsolvable pixels. In our method, we propose the MSB alteration (MSBA) technique, which modifies the MSBs of the red and blue channels to generate authentication codes with minimal distortion. This technique effectively reduces image distortion and the number of unsolvable pixels. Furthermore, we introduce the pixel substitution strategy (PSS) and the constraint-reduced strategy (CRS) to enhance the embedding efficiency. Experimental results demonstrate that our method outperforms prior work by converting more unsolvable pixels to solvable ones and achieving better detection results. Moreover, our method enhances image quality by approximately 1.5 dB, while maintaining the same embedding capacity.