Multilevel DTCWT for feature extraction and SVD image registration for micro air vehicle autonomous navigation

In this paper, image registration of aerial image with reference satellite image is carried out by computing multi-level sub bands using dual tree complex wavelets. Features are extracted from sub bands and the corresponding features are mapped for similarity by SVD methods. The registered images are evaluated for their performance metrics and the improvements in registration process in terms of transformational parameters are evaluated. The registration process proposed is accurate as it identifies the features with six orientations and thus compares the features and performs orientations foe perfect registration. The algorithm is designed to be faster by eliminating false features thus making it suitable for real time applications.