image - How we decide the region of computing SIFT Descriptor? -
a keypoint descriptor created first computing gradient magnitude , orientation @ each image sample point in region around keypoint location. confused meaning of "region". size of it? depends on keypoint's magnitude ? or kernel size of gaussian in scale level keypoint found ? or else ?
it depends on "kernel size of gaussian in scale level" - in sense "what region original image take". depends on sift parameter - patchsize (usually=41) size window in pixels reproject region original image.
Comments
Post a Comment