close
1.feature descriptor : The intent of a feature descriptor is to generalize the object in such a way that the same object (in this case a person) produces as close as possible to the same feature descriptor when viewed under different conditions.

2.HOG(Histograms of Oriented Gradients) person detector 較簡單,因為使用的是global feature而不是一堆local feature的集合(一個人可以用single feature vector就可表示,而不是用many feature vectors representing smaller parts of the person)

3. gradient vector: 針對每個像素a,將像素a左邊的像素b和右邊的像素c相減(要b-c或c-b都可,但整張圖所有像素減的方向要一致),得到x;y方向同理得到y,(x,y)可用來算tan值(角度)和大小(平方相加開根號),由此可得到一個向量
原教學: 
https://chrisjmccormick.wordpress.com/2013/05/07/gradient-vectors/     (英文,不過講解超清楚!!!)
gradient vector用途: (1)edge detection (2)feature extraction(可避免掉lighting對影像造成的影響) 

資料來源: https://chrisjmccormick.wordpress.com/2013/05/09/hog-person-detector-tutorial/

arrow
arrow
    全站熱搜

    慈 發表在 痞客邦 留言(0) 人氣()