Optical flow using textures

Generalized Optical Flow Model for Scattering Media

“…Different constraints can be assumed such as: Global smoothness [1], local smoothness [18], texture constancy [19], and multiple color channel [20].…”

“…In order to tackle the scattering and absorption affects in underwater images to estimate the optical flow, [19] proposed a method that combines intensities and texture. Despite the good results in real underwater images, these were shown only for images acquired from scenes near to the camera.…”

Fast modified Horn & Schunck method for the estimation of optical flow fields

“…It is simple and can be implemented in real time that produces good estimates except at motion boundaries. Several approaches have addressed this topic and proposed methods that account for velocity boundaries at the cost of significant computa-tional complexity, which makes them inadequate for current real-time applications [1][2][3][4].Another problem with this algorithm is that the intensity derivatives are estimated by using a first-order difference , which is a relatively crude form of numerical differentiation. This problem can be solved by using a spatiotempetal Gaussian prefilter with a standard deviation of 1.5 pixels in space and 1.5 frames in time (1.5 pixel-frames), sampled out to three standard deviations.…”

Colour Constrained 4D Flow

“…This extra information can further constrain and thereby improve the accuracy, reduce aperture ambiguities, and increase density in the flow estimates -as shown in the case of incorporating intensity [10]. Research in further constraining optical-flow techniques has also investigated the use of texture [2] and colour [1]. Relatively little work has applied these potential sources to range data despite their obvious presence in vision, and in applying the benefits to 4D flow.…”