Based on energy distribution analysis oforthogonal FRIT coefficients, a novel digital imagewatermark embedding and blind detecting algorithm inridgelet domain is proposed in this paper. Since the ridgelettransform has directional sensitivity and anisotropy, theimage is first partitioned into small pieces and theorthogonal FRIT is applied for each piece to obtain a sparserepresentation of the image, especially for straight edgesingularity. Through analyzing texture distribution inridgelet coefficients of each piece, these image pieces areclassified into frat regions and texture regions by Fuzzy CMean(FCM) clustering algorithm. The texture regions usedfor watermark embedding also classified into two regions,strong texture regions and week texture regions for differentembedding strengths based on the features of luminancemasking and texture masking. And the watermarks can beblindly detected without the original image and watermarkinformation. Experimental results show that the proposedwatermarking scheme can achieve a better tradeoff betweenthe robustness and the transparency.
展开▼
