Abstract:

The discrete cosine transform based steganography has become an important method of ensuring the safety of data transmission, because discrete cosine transform based steganography is less vulnerable to compression attacks compared to other steganography techniques. In this paper, a suggestion is made to a new steganographic method that relies on parity-based alteration of the mid-frequency DCT coefficients of carefully chosen coefficients to effectively allow the maximum amount of embedding and the capability to endure. The suggested approach is an intelligent scheme of selecting coefficients that selects appropriate mid-frequency components, and has an adaptive scheme of parity modification in regards to local image characteristics. The experimental performance is better when the PSNR is above 42 dB, SSIM score is above 0.985 and extraction accuracy is 99.2 percent even in JPEG compression with quality factors up to 70 percent. Its undetectability is verified through comprehensive security analysis versus state-of-the-art steganalysis techniques and the efficiency of its computation is analyzed with regard to performance of O(n log n), which is suitable in real time applications. The embedding capacity of the proposed technique is up to 0.52 bits per pixel, whereas visual quality is also similar to that of the original image and the method is more robust, imperceptible, and secure compared to the currently used DCTbased steganographic techniques.