Two-band Infrared Data Fusion Method Based-on Fractal Dimension

Researches indicate that the gray-scale images mapped from most nature objects accord to the fractal Brown stochastic field which has a foundation of self-similarity, meaning the image is made up of copies of itself in a reduced scale. The fractal dimension can quantifiably depict the fractal character and the property of an image. Therefore, a new method based-on fractal dimension to fuse mid-wave infrared images and long-wave infrared images is presented in the paper. First, images are decomposed into different scales using wavelet transform. Second, the fractal dimensions of sub-images of original images are computed, and then weighted to fuse sub-images at the same level according to different direction. Finally, the fused images are reconstructed by wavelet inverse transformation. The experimental results imply that the method can effectively preserve the information of source images with a high contrast and may be very practical