Study On IR/Visible Image Registration

The infrared and visible images are very different in grey level and image features according to a same scene, which make it difficult to design proper registration algorithms and to ensure the precision between them. The registration algorithm which is combining the mutual information-based image registration and edge feature-based image registration to images who have obvious edge features can implement simply and can improve the robust of the registration processing, since it can make full use of the edge features so that the differences of the gray level and the features of the image has been reduced. The compute-cost of mutual information is too large, so we bring forward a new algorithm based on wavelet. This algorithm speed registration process through matching images layer by layer which is obtained by wavelet analysis technology.Experimental results show that our combined algorithm solve the registration problem with infrared and visible images which have great differences in grey level and image features, and make registration time one magnitude faster than traditional algorithms due to the adoption of the wavelet transform. Simultaneously, the registration results achieve a high accuracy and stability.