Hiding text using the least significant bit technique to improve cover image in the steganography system

Abstract

One of the highest priorities in the era of information technology is to achieve an accurate and effective system for hiding security data. One of the goals of steganography is imperceptability to intruder. So this paper work to increase the imperceptibility on image, which has weaknesses in previous studies, as well as to avoid statistical attacks such as chi-square. A method has been proposed that includes calculating the color contrasts in the homogeneous areas of the image and dividing them according to the color contrast and exploiting the data of pixels that have a high impact to embed on the two first and third bits of least significant bit (LSB) to increase the amount of embedded data, impact regions (IR) classify according to selected features extracted in advance by using the support vector machine (SVM) classifier. Work was done on standard images taken from a standard dataset (USC-SIPI) for two types of gray and color images. The results showed the worth of the proposed method through a high peak signal to noise ratio (PSNR) that reached 89.5 dB due to the distribution of data on pixels according to the proposed method.