Image Fusion Of High-resolution And Multi-spectral Based On Multiwavelet Transform Research

With various kinds of remote sensing satellite sensor emerging,Many High-resolution and Multi-spectral images are available.These sensors have their own characteristics with different constraints and design.In order to take advantage of these characteristics,to make up for information loss of single image and enhance image information analysis capabilities,image fusion is coming. Remote sensing image fusion is a comprehensive multiple discipline.The paper is based on the pixel level fusion to study how to take advantage of these characteristics in High-resolution and Multi-spectral images.Fusing image is keeping light spectrum of Multi-spectral images as much as possible with the detail characteristic of space in the High-resolution image.Firstly to introduce basic method of remote sensing image fusion,evaluation method of the final fusing image and image registration with pixel-level image fusion, then these methods used in image fusion and experiment comparison accordingly. After introducing the wavelet transforming method used widely, however, it is not perfect. With the introduction of multiwavelet transform as the expansion of wavelet transform, multiwavelet transform has excellent properties of symmetry, orthogonality, short support and two vanishing moments.So the image processing with multiwavelet transforming has advantages over wavelet transform.Studing the characteristics of multiwavelet transforming and finding its unique advantages.So GHM wavelet transforming is used to analyze the image,and more detailed information of high frequency and low frequency information(in all directions)is got.Because of the multiple wavelet analysis of image using a vector decomposition, the decomposed frequency information should comply with IHS decomposition characteristics,the method of vector fusion to keep the Multispectral information.IHS is in line with the human vision.RGB color space is often used to express image,but it does not represent the perception of the human eye, so from the beginning of the image fusion, IHS transforming to transform RGB color image into IHS space to separate the brightness(I), hue(H) and saturation(S) information for further processing.Pulse coupled neural network(PCNN) derived from biology, and is different from the traditional neural network structure, the paper studied the characteristics of PCNN, adopting simplified PCNN model as a way of high frequency information fusing. The link intensity ofNeurons model used in the article, is an adaptive number according to image direction information.Therefore the main research method of the paper is IHS transformation combines with wavelet transform and PCNN image fusion method.