[ptx] Question on BBF used in Brown's paper.

[Xianyong Fang]

Hi Sebastian,

	Thank you very much for your detailed explanation. I really hope that my BBF algorithm works. As you said that these two algorithm are the same, I will try to use the priority algorithm in ANN directly. I am not confident with my programming skills and ANN must be faster than my code. 
   
    I also appreciate your nice remindess that the ration of the first and second hit is used to decided whether a keypoint has matches. I omit this features actually,:)

Xianyong


>So, just plan beforehand and not just finish each feature alone. :)
>
>For the k-nearest neighbours I found the BBF performance to be decreasing until
>it found k neighbours. But for the keypoint matching it is very helpful to have
>the ratio of the first and second hit, as Brown and Lowe recommend.
>
>
>> Xianyong
>
>Zaijian,
>Sebastian
>
>-- 
>nowozin at cs.tu-berlin.de --- http://user.cs.tu-berlin.de/~nowozin/
>

= = = = = = = = = = = = = = = = = = = =