Study Of Multisensor Image Fusion Algorithms

Image fusion,which has been an important and useful technique for image analysis and computer vision in recent years, is a technique to combine information from multiple images of the same scene.The methods and rules of image fusion based on multiresolution are discussed in this thesis. The multiresolution image fusion technique based on Gauss-Laplacian pyramid decomposition is studied in detail. Region-based measure fusion rule, energy method and maximize standard deviation, are applied to fusion experiments. Two decomposition methods ration pyramid and contrast pyramid are introduced too.Wavelet analysis has the advantages of multiscale, multiresolution and nonredundant. An image fusion schema, which based on region's characteristic in wavelet domain, is introduced in the paper. It bases on two characteristics: the human visual system is sensitive to local luminance contrast, and the other one is that some region is contains more information when this region has biggish standard deviation. A series of experiments on image fusion show that this scheme has preferable fusion result.Quantitatively assessment of the image fusion performance is an important and a complicated issue. The objective evaluation of image fusion performance is studied. A new evaluation criterion is presented in this thesis.In this dissertation, IHS transform and fusion schemes based on IHS and wavelet transform are discussed. A new fusion algorithm based on lαβand wavelet transform is proposed. The algorithm is introduced into the field of weapon detection and the result is well. The classical false color fusion schemes and color transfer algorithms are discussed in detail. Then a new false color fusion scheme based on color transfer is given. Experimental results show that the algorithm is effective.At last, the schemes and current situation of image registration are introduced.