Center-Symmetric Local Binary Pattern-Based Image Authentication Using Local and Global Features Vector

Abstract

Center-symmetric local binary pattern (CS-LBP), an important tool for feature extraction, which is widely used for object recognition. In this research work, a reversible watermarking technique has been introduced for image authentication making use of local and global feature vector generated by exploiting CS-LBP, discrete cosine transform (DCT) and discrete wavelet transform (DWT). In the embedding procedure, authentication code (AC) is generated by employing CS-LBP operator on the colored cover image using a shared secret key and local feature vector and global feature vector generated by exploiting DCT and DWT which has been stored in an INFO matrix as well as embedded within the host image. In the detection procedure, authentication code (AC) is extracted with the help of a shared secret key and verified its originality by comparing it with a shared INFO matrix. Here, the symmetric property of LBP provides a minimum change in the watermarked image which helps to enhance the quality and authenticate the watermarked image.