| From the point view of mathematics, signal and image processing can be uniformly seen as signal processing. Now signal processing has become an important part of modem science and technology. Both of them are involved in complex analysis and specification of time series. In a word, a signal is a function of time. The aims of signal processing are correct analysis, effective code, speedy transfer and precision reconstruction.The classical signal analysis method is Fourier transformation, whose essence is to decompose the signal wave into many sine waves, and which also can be seen as representation of time function in frequency domain. Because Fourier analysis uses a global transformation, which all in time domain or all in frequency domain, it cannot figure out the local property time- frequency domain, which exactly is the most important property of unsteady signal. Fourier analysis has been changed to fit to analyze unsteady signal, which is represented by short-time Fourier transformation. But in essence, the short-time Fourier transformation is a method of ingle resolution because it uses an unchanged short-time widow function. Just for mat short-time Fourier transformation has fatal shortage in signal analysis.Fourier has been predicate mat any function (f (x) ) which has cycle of 2 n is the sum of his Fourier progression in 1870. But a cyclic function whose variable (x) has the cycle of 2 n has been constructed, which Fourier progression is radiation in some given point. There were three unresolved problems in that time, but all these problems led to important result. The three problems are:(1) To fit Fourier progression by modifying requirements of the function.(2) To modify the definition of Fourier progression.(3) To make trig-cluster constringed by acquiring other orthogonal base.The third problem brought on the birth of wavelet that has gone through about a hundred years. So wavelet transformation is the result of classic Fourier transformation's incompetence in signal processing.Wavelet transformation is signals' time梞easurement (time-frequency) analysis method. It has the characteristic of Multi-resolution Analysis and has ability of representation local property of time and frequency domain. Wavelet transformation is a kind of localized time-frequency analysis method with fixed size window of time or frequency with variable form. It has high resolution of frequency and low resolution of time in low frequency part This characteristic is fit to probe normal signal with instantaneous abnormal phenomenon and show its components. So, Wavelet transformation is seen assignal microscope. There is good effect to detect malfunction in dynamic system by means of continuous wavelet transformation. Wavelet package is division of wavelet high-frequency space for increasing resolution. By this means, the shortage of low frequency resolution on die premise of high time resolution can be overcome.Due to the future of wavelet transform is accordant with the human visual model (HVS), a watermark can be embedded into an image by using wavelet transform, and its covered effect is apparently superior to the one inset by using other way. A method of inserting a watermark into a static image, which is based on the technique of multi-resolution wavelet analysis, is suggested in this dissertation. The innovative idea of the method is in that the analytic ability for high-frequency signals of wavelet packet is used to enhance information capacity. Because the human vision is insensitive to high-frequency signals, one can make information of watermark mass and hightensile in the high-frequency domain. In this way, not only the quality of original image can be reserved, but also the robustness of the watermark improved. In the mean time, some experiments have been made to investigate the effect of the watermark, and die results confirmed its good concealment and excellent robustness.
|
PDF Full Text Request