HISTOGRAM BASED AUTOMATIC IMAGE REGISTRATION

Automatic image registration is a challenge in several fields. A novel approach of automatic image registration is presented in this work. Several methods for automatic image registration have been proposed in the last few years, it is still far from a broad use in several applications, such as in remote sensing. A method for automatic image registration through histogram-based image segmentation allows for a more detailed histogram-based segmentation, rather than the traditional methods, and consequently to an accurate image registration. The work enables to estimate the rotation and/or translation between two images, which may be multi-temporal or multisensory, with small differences in the spectral content. The application of this method is illustrated to simulated rotation and translation. The first dataset consists of a photograph and a rotated and shifted version of the same photograph, with different levels of added noise. It was also applied to a pair of satellite images with different spectral content and simulated translation, and to real remote sensing examples comprising different viewing angles, different acquisition dates and different sensors. Accuracy below 1 degree for rotation and at the sub pixel level for translation was obtained, for the most part of the considered situations. The main drawback of this method is the computational time, mainly associated to the segmentation stage. However, it is expected to optimize the implementation code in the future in order to provide a faster performance.