In the double-flow network, it is correct to say that the optical flow method is easily influenced by illumination factors.

Optical flow is a concept in object motion detection in the visual field. Used to describe the movement caused by the movement of the observation target, surface or edge relative to the observer.

Optical flow method is very useful in pattern recognition, computer vision and other image processing fields, which can be used for motion detection, object cutting, calculation of collision time and object expansion, motion compensation coding or stereo measurement of object surface and edge.

1950, Gibson first proposed the optical flow method. When people's eyes observe a moving object, the scene of the object will form a series of continuously changing images on the retina of the human eye. This series of ever-changing information constantly "flows" in the retina (that is, like the plane), just like the "flow" of light, so it is called optical flow.

Its basic idea is to describe the motion of the target in space in the form of a sports field, that is, to give each pixel in a frame of image a velocity vector.

Optical flow calculation technology;

Region matching method. In the region matching method, optical flow is defined as the best fitting displacement between image regions at different times. Region matching is to maximize similarity measures such as SSD, mutual information or correlation coefficient to achieve the optimal matching of regions.

Energy-based approach. Energy-based method is the mainstream algorithm for optical flow calculation at present. Its basic idea is to transform optical flow calculation into a global energy function optimization problem with a series of constraints. Usually, the constrained optimization problem is further transformed into unconstrained optimization problem by penalty function method.

Phase-based method. Optical flow can be determined by the phase characteristics of the band-pass filter output, so it is called phase method. The phase model is actually to transform the problem into frequency domain.

At present, optical flow method is widely used in many important computer vision and image processing fields, such as target segmentation, recognition, tracking, robot navigation, target shape information recovery, three-dimensional structure recovery and motion estimation. In addition, it is also widely used in medicine, oceanography, astronomy and other related fields.