A central idea in the use of multiresolution techniques in image processing is the analysis of local regions of different sizes and the determination of some form of optimal decomposition in terms of such regions. The idea underlies many of the multiresolution approaches used in image segmentation, for example. However, similar schemes can also be applied to other analysis tasks. Two related examples are binocular disparity estimation for stereopsis and motion estimation from image sequences. The goal in both cases is to identify corresponding regions in two images and in each case the correspondence will inevitably involve regions of different sizes. The purpose of this paper is to outline a multiresolution approach to finding local correspondence in images which can be used in both disparity and motion estimation. The scheme is based on performing local correlations between regions via the frequency domain and employs a course-to-fine matching strategy to determine an `optimal' correspondence decomposition. The scheme is implemented within the framework of a generalised wavelet transform, the multiresolution Fourier transform (MFT), which also provides the potential for incorporating into the scheme both feature information and more complex matching criteria, such as those based on affine transformation. Results of experiments illustrating the effectiveness of the approach in both applications will be presented.