Distorted pattern fusion using wavelet transform

An effective pattern fusion method based on wavelet transform is proposed. After geometric change, corrupt pattern images are obtained from distorted pictures. Considering the special characters of the pattern or text, a novel fusion scheme is then developed to de-noise, register and fuse in order to get a lucid one for further discrimination. The scheme is based on wavelet transform. The efficiency of this new method is verified by experimental results. To our knowledge, it is the first attempt to consider the pattern registration and fusion.