An Evaluation Of Multiwavelet Families For Stereo Correspondence Matching

This paper presents an evaluation of different types and families of multiwavelets in stereo correspondence matching. Different multiwavelet families with different filter types such as balanced versus unbalanced, symmetricsymmetric versus symmetric-antisymmetric are used. Normalized cross correlation is employed to find the best correspondence points and generate a disparity map. In the case of balanced multiwavelets, due to similar spectral content of the four generated low frequency subbands, they are shuffled to form a single baseband and then this baseband is used to generate a disparity map. However, in the case of unbalanced multiwavelets, the resulting basebands are used to form four disparity maps and then these maps are combined using a Fuzzy algorithm to generate a single disparity map. Middlebury stereo test images are used to generate experimental results. Results show that the unbalanced multiwavelets produce a smoother disparity map with less mismatch errors compared to balanced multiwavelets. Keywords-Multiwavelets; stereo correspondence matching; normalized cross correlation;