GLCM and PCA algorithm based watermarking scheme

• Published in: AIP conference proceedings Vol. 2916; no. 1
• Main Authors: Dewan, Sakshi Garg, Kalra, Mala
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Melville American Institute of Physics 05.12.2023
• Subjects: Algorithms; Audio data; Bit error rate; Correlation coefficients; Digital imaging; Digital watermarking; Digital watermarks; Discrete Wavelet Transform; Embedding; Feature extraction; Image quality; Machine learning; Performance evaluation; Signal to noise ratio; Texture; Wavelet transforms
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0177800

Digital watermarking is a way to protect the data from illegal access. Digital watermarking is method of hiding secret information of image, audio, video, text form within same form. Image watermarking is embedding an image inside an image. This research work suggests a novel framework for Image Watermarking Technique to achieve better image quality. In this work, Discrete Wavelet Transform (DWT), Gray Level Co-occurrence Matrix (GLCM) and Recursive Feature Elimination (RFE) are used for watermark embedding and extraction. The watermark is embedded into an original image with Online Sequential Extended Learning Machine Algorithm (OS-ELM). By combining GLCM for texture feature extraction and RFE for feature selection, the watermarking system can benefit from the discriminative power of texture features while ensuring that only the most relevant and informative features are utilized. This approach helps in improving the robustness, security, and perceptual quality of the watermarking system by focusing on the most discriminative and watermark-relevant texture features. During extraction process, the wavelet and textural features of the watermarked image are extracted and input to the Inverse OS-ELM to extract the watermark. The performance evaluation of the proposed work is done against the existing work on the benchmark of various metrics namely, Peak Signal to Noise Ratio (PSNR), Mean Squared Error (MSE), Bit Error Rate (BER), and Correlation Coefficient. It is evident from the simulation outcomes in MATLAB, the proposed work outperforms the competitive algorithms.