Watermarking Algorithm And Edge Detection Based On Wavelet Transform

Watermarking algorithm and edge detection base on wavelet transform are investigated in this thesis. First of all, importance as well as domestic and international development of digital watermarking and edge detection are introduced. As the basic theoretics of continue wavelet and discrete wavelet, multi-resolution analysis, two-dimension wavelet transform and image manipulation are given. Concretely, three aspects are finished in this thesis.First, using chaotic sequences, an algorithm of embedding watermark in low frequency subband of wavelet transform image is designed. But the low frequency subband is not suitable for embedding because of the degraded image quality. In order to overcome the problem, chaotic sequences are embedded into insensitive locations. The quality of algorithm is tested on the Lena image of 256×256×8bits, the results of experiment show that the watermarks generated with the proposed are invisible and the method is robust enough to some image processing techniques, such as noise, JPEG compression, dither and so on. So it is effective.Second, using wavelet transform, multi-resolution and maximum modular, an algorithm for detecting the edge of steel ball image is advised. However, it is difficult to detect the edge of steel ball for unartificial noise in the image. In virtue of wavelet transform, the gray change of image is known and the disturbances of noise are voided. The results of experiment show that the complex calculation defect of Canny algorithm is overcame, the noise effect is removed, and the nicks on the steel ball are visible.Third, the discrete smooth wavelet transform is applied in oil data analysis. In oil data analysis, for the existence of noise and the effort of singularity points, indention distribution is appeared, and normal analysis work is affected. For the redundancy characters of discrete smooth wavelet transform, mid-way threshold method is applied to remove the noise disturbance and smooth the whole procedure data. Furthermore, redressal coefficient can be elected for different states and the request of engineering technology for measure data is satisfied.