In the field of medical ultrasound imagery, the knowledge of displacement is an important input for many clinical exams, including color Doppler and elastography. In the literature, the two main way proposed for solving the displacement field estimation are blockmatching techniques and Doppler methods. Blockmatching techniques try to match part of images acquired before and after the displacement and depends on the way those images where created. On the other hand, Doppler methods only estimate displacement along a fixed direction. In this thesis, we proposed three methods that can be used to estimates directly the motion field from raw RF signals by using the proposed concept of normal vector. This concept models how displacement will change RF signals. The proposed methods are compared in simulation and experimentally to a reference blockmatching technique and show a improvement of resolution. Proposed methods resolution are up to 1.5µm in the transverse direction and 2.5µm in the axial direction at a frequency of 5MHz. We also proposed a new beamforming technique that estimate the spectrum of the ultrasound image. This method is compatible with RF signals used by motion estimation methods.
