A Study On Enhancement Processing Of Medical Images Based On Dyadic Wavelet Transform

It is necessary for enhancement of medical images what have poor contrast and unclear details to meet diagnosis and treat. Discrete wavelet transform has received much attention as a new powerful tool of image processing. There are two types of discrete wavelet transform: the biorthogonal wavelet transform and the dyadic wavelet transform. Since the biorthogonal wavelet transform is of down-sampling type and the demerit of being not shift-invariant, so precision of image processing isn't very good. However, the dyadic wavelet transform is shift-invariant and don't sample, so precision of image processing is better. Now, we study on enhancement of medical images based on the dyadic wavelet transform theory. Firstly, We analyse the dyadic lifting schemes what are proposed by T.Abdukirim. The schemes generalize Sweldens lifting schemes. Employing the schemes, biorthogonal wavelet filters and dyadic wavelet filters with good properties can be constructed starting from dyadic wavelet filters by choosing appropriate free parameters. In thesis two theorems are proposed based on the dyadic lifting schemes and vanishing moments. Two theorems offer a new method of construction of dyadic wavelet filters. Two theorems indicate that dyadic wavelet filters with higher number of vanishing moments can be obtained if free parameters meet condition of theorems. Meantime, lifting dyadic wavelet filters with compactly supported, symmetry and higher number of vanishing moments are constructed starting from initial dyadic wavelet filters constructed by T.Abdukirim. Secondly, initial dyadic wavelet filters and lifting dyadic wavelet filters are used in experiments. Results of lots of experiments show that Entropy and Contrast Improvement Index of images processed by lifting dyadic wavelet filters have slighter improvement compared with initial dyadic wavelet filters. So it is necessary and valid for construction of lifting dyadic wavelet filters. Medical images what have higher contrast and clearer details could be obtained. It is in favor of assisting doctors in diagnosising, and offers reference for development of radiology.