图像特征点匹配（Matching feature points） to be continued..

LIFT应该没有开源匹配的源码，今早霍老师要求做一下匹配，那就做一下吧，顺便学一学，争取今晚给他一个结果。

优达学城上计算机视觉概论讲到了Matching feature points（在SIFT讲完之后），所以下面先把视频中提到的内容归纳一下：
1. Exhaustive Nearest Neighbor（computationally expensive）
2. Approximate Nearest Neighbor（better than the previous one）

• Best-bin-first algorithm used in SIFT(Beis & Lowe, 97), it’s a modification to k-D tree algorithm
• Use heap data structure to identify bins in order by their distance from query point

Result: Can give speedup by factor of 100-1000 while finding nearest neighbor (of interest) 95% of the time
3. Another cool thing you can do is do some form of a hashing based method in order to index. eg: Wavelet-based hashing(compute a short (3-vector) descriptor from the neighborhood using a Haar “Wavelet”)

4.Locality Sensitive Hashing, idea: contruct hash function, then we can solve the problem by hashing

后面两个还没搞太懂，尤其是Wavelet-based hashing

后面还有3D Object Recognition等内容，以后再看吧。