Image Digital Watermarking Based On Wavelet Packet With Best Tree

The technique of wavelet packet transformation has been applied in all kinds of fields in recent years. Wavelet packet can analyze the high frequency part of signal, developing room of watermarking information embedded consumedly, and this method of embedding watermarking information can applied to encryption of image. So, the method of watermarking embedding with wavelet packed transformation has got more and more attention of people. At the present time, more and more methods of watermarking embedding with wavelet packed have appeared in grey image and color image, obtaining more moderate watermarking. But these methods have many limitations. Some methods have no best basis, some methods have perplexing arithmetic of finding the best bases, and others have a great deal of data to deal with.This paper studied the technique of image watermarking with wavelet packet, aiming at all kinds of limitations of wavelet packet watermarking, bringing forward a more effective method of embedding and extracting, and comparing to the existing methods. We have proved the feasibility and validity of our method with many preparations. All our work and productions are listed as follows:1 , thinking of the peculiarity and infections to watermarking of image, in the wavelet packet transformation of the original image, we chose the feasible basis called daubechies 7/9 wavelet base and use the entropy-based algorithm to select the best wavelet basis. This entropy-based algorithm which can stand against the effect of noise in compression is usually used in image compression. This is just be required in watermarking. So this basis has potential in processing image watermarking.2, Embedding watermarking in the low frequency of image can bring biggish effect and the watermark can be easily identified. The high frequency has noise and most methods of compressing is to wipe off the high frequency of the signal.If we limit the partition of frequency to a small range, and this partition has a frequency block called interesting block having high energy and can not be easily identified, this frequency block can not be affected by the operation of compressing. Our paper just select this kind of block to embed watermarking.3^ We compare our method with multi-wavelet watermarking which is new in watermarking field. We compare them in basal theory and process and do many attack experiments to prove the validity of our idea. The results are also be summed up and analysed.