Registro de Imagens Multitemporais Baseadas em Decomposição Multiresolução

Resumo

Image registration is the process of overlaying two or more images of the same scene taken at different times, from different viewpoints, and/or different sensors. With the increase in the number of images collected every day from different sensors, automated registration of multi-sensor, multi-spectral and multi-temporal images has become an important application. In this paper an efficient algorithm for the registration of multi-temporal images with similar spectral responses is proposed. The procedure relies on the gray level information content of the images and their local wavelet transform modulus maxima. The registration algorithm is performed at progressively higher resolutions, which allows for faster implementation and higher registration precision. For the matching process, a novel procedure for automatic point feature detection based on wavelet transform is presented. Moreover, an investigation about the performance of four similarity measures in the task of feature point matching for satellite images is also presented. The results show that the matching method based on correlation is efficient to register images of similar spectral responses even if multi-temporal changes are present. To overcome the presence of rotation angle between the images to be registered, a procedure to estimate this angle is performed before the matching process. Measurements of performance are provided, as well as extensive experimental results, to illustrate the qualitative performance of the registration algorithm. Images of different sensors and acquisition time are used in the experiments.