Image Fusion Method Research Based On2D Non-separable All Phase Directional Filter Bank

With the continuous development of image sensor technology and computer technology, the application of image fusion technology is becoming wider and wider. It can be used in the fields of military affairs, medical science, remote sense, target identification, machine vision. For now, tensor product wavelet transform in the image fusion has been widely used. But it is neither fully using data itself unique features, nor the optimal function representation method. It can catch only limited direction information of horizontal, vertical and diagonal directions, so that it is not fully using image itself geometry regularity. Recently, with the continuous research of image directions, several transforms have been proposed, which are ridgelet, curvelet and contourlet and so on. These transforms have flexible direction selectivity and anisotropy, and the have become outstanding in the study of image directions. But the directional filter bank of Contourlet transform has some disadvantages.All phase directional filter bank (APDFB) produced by the previous people can make up the loss of direction brought by traditional tensor product wavelet and overcome the insufficiency of directional filter bank of Contourlet transformation, it is used in image fusion. The APDFB is a true2dimension direction filter bank. It gets rid of the down-sampling and rotation operation of images, it reduces computational complexity and reserves detail information of image direction. It also has good direction selectivity, and don’t need reconstruction filter when reverting, it only need to add different directional information. The APDFB has many characteristic, which are translation invariance, more directional and anisotropy, multiresolution analysis (MRA) and so on. The dissertation used the APDFB in the image fusion, and made the following aspects.On the one hand, combined between the laplacian pyramid transformation and the all phase directional filter bank transformation, a multi-spectral image fusion method using2D contourlet transform is presented. Firstly, the original multi-spectral image is converted from RGB mode to IHS mode through IHS transform. Next, the intensity component and the panchromatic image are merged into a new intensity component according to the fusion rule of selecting the pixel that the region energy is big based on the2D contourlet transform. Finally, the final RGB image is achieved by an inverse IHS transform with the H, S and the new intensity component. The experimental results show that, using the same fusion method, the fused image using the2D contourlet transform has better quality than using contourlet transform and wavelet transform. Also, the method improves fusion performance over the IHS fusion method in spectral characteristic.On the other hand, combined between the wavelet transformation and the all phase directional filter bank transformation, a multi-focus image fusion algorithm using wavelet based all phase directional filter bank transform is proposed. First the wavelet based all phase directional filter bank transformation to the different focusing image was made, then the weighted averaging operator was used to carry on the fusion to the low frequency coefficient, the local variance criterion was used to carry on the fusion to each direction high frequency coefficient. Finally, to reconstruct the coefficient of fusion, the fusion image in which the high and low frequency proportion elevates was obtained. The experiment results show that, the method significantly reduces the RMSE value, improves the PSNR value, and makes that the fused image has good effect.