Quaternion based assessment method for color image fusion algorithms

A new approach to assess the performance of color image fusion algorithms is proposed. The color information is fully used by this method. Quaternions are used to encode the pixels in a color image into a quaternion matrix. Local variance of the luminance layer of the color image is taken as the real part of a quaternion, then the three RGB channels of the color image are encoded into the three imaginary parts of a quaternion. The angle between the singular value feature vectors of the quaternion matrices corresponding to one of the source image and the fused image is used to measure their structural similarity. Edge information transferring is used to balance the sensitivity of the assessment methods. Experimental results show that the proposed assessment method is better consistent with the HVS than some of the state-of-the-art method. Accurate assessment results can be given by the proposed method for various of color image fusion algorithms by using the source images and the fused image.