Performances evaluation of image fusion techniques based on nonlinear correlation measurement

In this paper we firstly propose the concept of nonlinear correlation coefficient, which is sensitive to the general relationship between two variables. Based on that, nonlinear correlation information entropy is developed to measure the general relationship among a group of images. A performances evaluation method based on nonlinear correlation information entropy is developed. Furthermore, the performances of two widely used image fusion techniques, i.e. wavelet transform based fusion and pyramid transform based fusion operating on a typical hyperspectral image set, are evaluated using the proposed method. And the performances evaluation results are agreed with the comparison results of classification accuracy in real application, which further prove the correctness and effectiveness of the proposed measure.