The Research Of Image Enhancement In Wavelet Domain

The image enhancement whose purpose is improving the visual perception is one of three basic tasks in digital image processing. The spatial domain method and the transform domain method are two board categories of the traditional image enhancement algorithms. In the spatial domain method, the pixels are manipulated directly. In the transform domain method, the transform coefficients of the pixels are manipulated. There are two kinds of transform used widely: Fourier transform and the Wavelet transform. The developments of the wavelet transform are very quickly because its mathematic though is very elegant and it is used widely in engineering field. The algorithm and application of the image enhancement based on the wavelet transform is discussed seldom although the wavelet transform is used widely in the image compression and segmentation.Multi-level hierarchical wavelet decompositions are used to image processing. The high frequency coefficient bands are primarily noise and detail. It can't cause very serious effect when the low frequency coefficient bands are smoothed. A wavelet threshold de-nosing algorithm based on the pixels classification is presented in this paper. The low frequency coefficient bands are smoothed; the noise and detail are divided by the pixels classification. The de-noised image is reconstructed which the noise coefficients are set zero.The wavelet transform of the ramp edge is its one order derivative actually. It is shown as the roof edge in the high frequency coefficient bands of wavelet transform. The contour which lies at the orthogonal edge section approximates Gaussian or raised cosine function probability distribution. The edge is enhanced by means of changing the contour of Gaussian or raised cosine function and getting the edge whose slope is change by means of inverse transform.The cubic B spline wavelet transform used, the image de-noising and edge enhancement is completed simultaneously. This is an effective method.