Fusion algorithm for multi‐sensor images based on PCA and lifting wavelet transformation

Abstract

Anovel fast image fusion scheme based on principal component analysis (PCA) and lifting wavelet transformation (LWT) is proposed. Firstly, the principal component images of the registered original colour image are obtainedby PCA transformation. Then, the first principal component image and near infrared imagery are merged using lifting wavelet transformation (LWT) based on regional features. The fused image replaces the first principal component of the visual colour image. Finally, the final composite image is obtained by inverse PCA transformation. Compared with other fusing algorithms, the experimental results demonstrate that this fusion scheme is more effective in fusing image quality than the traditional PCA or wavelet transformation fusion methods. The obtained image conforms to human vision features. The standard deviation (ó) and average gradients (g) are a little smaller with this fusion algorithm than the wavelet transformation method, but they are bigger with this fusion algorithm than the PCA method; however, entropy (EN) and correlation coefficients are larger with this fusion algorithm than with the PCA or wavelet transformation method. The fusion image contains more information and stronger spatial detail performance. The merged image is more advantageous to be further analysed, understood and recognised.