This paper presents a robust digital image watermarking scheme that is both geometrically invariant and relies on advanced techniques of feature extraction and local Zernike transforms. The methodology incorporates the Adaptive Harris Detector, a sophisticated algorithm that is utilized to identify and extract distinctive feature patches from the original image which are crucial for the watermarking process. The core of the proposed watermarking scheme lies in the application of local Zernike moments. This approach allows for the watermark to be embedded directly into the extracted patches through an innovative process that employs the inverse Zernike Transform. Each selected circular patch undergoes further decomposition into a collection of binary patches, which serves to enhance the granularity of the watermarking process. For each of these binary patches, Zernike transform is applied, enabling the capture of essential characteristics that are reflective of the original image’s features. The next step involves calculating the magnitudes of the local Zernike moments derived from the transformed patches. These magnitudes undergo a modification process to effectively encode the watermark into the image. Once the watermark is embedded, the inverse Zernike transform is utilized to reconstruct the modified binary patches. This reconstruction ensures that the visual integrity of the original image is maintained while the watermark remains securely embedded. Experimental results conducted in various scenarios demonstrate that the proposed watermarking scheme exhibits exceptional robustness against various types of geometric distortions, including but not limited to rotation, scaling, cropping, and affine transformations. Additionally, the method shows significant resilience against common signal processing attacks, further validating its effectiveness in real-world applications.
Geometric Invariant, Feature Extraction, Adaptive Harris Detector, Local Zernike Transform, Inverse Zernike Transform