computer vision - Advantages of SIFT over HOG -
i studying sift (scale-invariant feature transform) , hog (histogram of oriented gradient). want know advantages of sift on hog in case when have extract features sequence of frames of activity (i.e. standing). until have understood:
1) in sift gaussian smoothing applied in order compute dog (difference of gaussian). performing scale extrema detection detect feature points. once have feature points need compute hog each feature. since takes 16x16 neighbourhood result 128 length descriptor. whereas hog compute edge gradient of whole image , find orientation of each pixel can generate histogram.
2) hog used extract global feature whereas sift used extracting local features.
3) sift scale , rotation invariant whereas hog not scale , rotation invariant.
please let me know if there other advantages of sift. correct me if have wrong understanding. thank you
Comments
Post a Comment