Determination of horizontal motion through optical flow computations

Accurate navigation (localization) is one of the most essential capacities for an intelligent/autonomous subsea vehicle. For many applications, reliable short-range horizontal positioning is difficult to achieve, particularly over flat bottom topography. A potential solution to this problem proposed in this paper utilizes a passive optical sensing method (optical flow computation) to estimate the vehicle displacement using the bottom surface texture. Fundamentally, the proposed method is similar to correlation methods. However, it is computationally more efficient than correlation methods since it avoids blind searching a large neighbor region. The proposed method does not require feature correspondences in images and is robust in allowing brightness changes between image frames