Determining dense velocity fields for fluid images based on affine motion.

Area of Science:

• Computer Vision
• Fluid Dynamics
• Image Analysis

Background:

• Traditional optical flow methods struggle with fluid dynamics due to deformation and lack of distinct features.
• Accurate fluid motion estimation is crucial for various scientific and engineering applications.

Purpose of the Study:

• To develop a robust and accurate method for estimating fluid flow between adjacent images.
• To overcome limitations of existing optical flow techniques in handling fluidic motion.

Main Methods:

• Utilizing an affine motion model for image patches to capture fluid motion.
• Implementing a best-buddies similarity approach for robust feature matching in fluidic images.
• Employing nearest-neighbor interpolation to create dense affine motion models and variational energy minimization for refinement.

Main Results:

• The proposed method successfully captures fluid motions using affine models and best-buddy similarity.
• Dense fluid flow estimation was achieved by applying affine transformations and variational refinement.
• Experimental validation across diverse fluid images demonstrated superior performance compared to traditional methods.

Conclusions:

• The novel approach provides a significant improvement in fluid flow estimation accuracy.
• The best-buddy similarity and affine motion model effectively address challenges posed by fluid dynamics.
• This method offers a reliable tool for analyzing fluid motion in image sequences.