比較好的論述文章 :
http://blog.csdn.net/abcjennifer/article/details/7639681
http://www.vlfeat.org/api/sift.html
1、高斯金字塔的創建 尺度連續性 冗餘金字塔
2、原作者推薦 第0個octave中的第0層 delta = 1.6 輸入圖像一般經過抗混疊濾波 認爲 delta = 0.5
3、方向參數的確定,如何設置窗的大小。
The magnitude and direction calculations for the gradient are done for every
pixel in a neighboring region around the keypoint in the Gaussian-blurred image L. An orientation histogram with 36 bins is formed, with each bin covering 10 degrees. Each sample in the neighboring window added to a histogram bin is weighted by its gradient
magnitude and by a Gaussian-weighted circular window with a that
is 1.5 times that of the scale of the keypoint.
高斯窗 1.5*delta
區域 3*1.5*delta
4、一張老被錯誤描述的圖形
其實這張圖只是描述了2*2*8 = 32 個描述符特徵
5、Vl-sift 對特徵抽取的描述
6、特徵點窗口大小如何計算?
從5中可以知道 binsize = m*delta
因此窗口矩形尺寸爲 m*delta*(d,d)
由於需要考慮旋轉的影響 m*delta* sqrt(2)*(d,d)
考慮插值 m*delta* sqrt(2)*(d+1,d+1)
半徑+0.5取整 m*delta* sqrt(2)*(d+1,d+1) /2+ 0.5