Abstract

This paper proposed a new robust digital watermarking technique with an aim of image protection. The proposed technique incorporates both the transform and feature methods. Initially, the Harris Laplacian detector extracts the features from the cover image that suites well for the watermarking. The extracted characteristic regions are grouped to form a primary feature set where the secret message is embedded to form a watermarked image. This image is tested by introducing six different predefined attacks, which checks the robustness of the extracted feature.Discrete cosine transform (DCT) transformation method extracts the secret message from the watermarked image. According to the results, the watermarked image is vulnerable to the undefined attack. To enhance the resistance against those attacks the primary feature set is extended by adding some auxiliary features. These features are extracted through Knapsack algorithm. The watermark is embedded into these auxiliary features where the secret message is not affected by any attacks. The experimental results show that the proposed watermark technique is efficient and also preserves the image quality.

Key words: Feature detector, Knapsack algorithm, watermark protection, attack resistant, digital watermarking.