Part Number: TDA2HG
SFM generally obtains 3D points by the parallax of multiple cameras.
TDA's SFM routine uses only one camera, and I guess the routine uses the difference between the frames before and after to get the feature points that need to be tracked. However, the angle of the surrounding scene during driving is relatively fixed, and the feature points of the preceding and following frames are not significantly changed.
I want to know how a single camera replaces the parallax technology of multiple cameras.
regards, chengbo
