Multi-scale image data fusion based on local deviation of wavelet transform

Abstracf-The goal of image fusion is to create new images that are more suitable for the purposes of human visual perception, object detection and target recognition. The use of multi-sensor data such as visible and infrared images has led to increased recognition rate in applications such as automatic target recognition. In order to adequately make use of all kinds of images information from multi-sensor, a new image fusion' method based on local deviation of wavelet transform is proposed. In the fusion processing, the fused approximate coefficients are obtained with weighted average method. For the bigger local deviation of the each decomposed approximate coeficient, we choose a big power gene. The other approximate metlident cheeses a small me. The fttsed detailed coefficients are obtained by setting each coefficient equal to the corresponding input image wavelet coefficient that has the greatest local deviation. Both the image information entropy and the image clarity can evaluate the performances of multi-sensor fusion algorithm. Some different methods are compared by computing their entropy and image clarity. The experimental results show tbat the image data fusion method based on local deviation of wavelet transform is more effective than the other methods.