What is the input to the featureXY parameter? I obtain features from fcvCornerFast9u8, but those features are of type uint32_t * __restrict. featureXY requires of type const float *. I am confused as to why featureXY is of type float * and not int. Does this imply there is a step between detection and matching that needs to be done, apart from casting?
fcvTrackLKOpticalFlowu8 Usage
Posted: Mon, 2016-01-25 20:32
Yes you need to convert the feature point coordinates to float type before using optical flow to track. You get fractional precision for the feature points although at start they are on integer grids.
Cheers,
-Jeff