opencv-api FlannBasedMatcher

Fast Library forApproximate Nearest Neighbors

1.创建对象

<FlannBasedMatcher object> = cv.FlannBasedMatcher( [, indexParams[, searchParams]] )

indexParams

参数	描述
FLANN_INDEX_LINEAR	线性暴力(brute-force)搜索
FLANN_INDEX_KDTREE	随机kd树，平行搜索。默认trees=4
FLANN_INDEX_KMEANS	层次k均值树。默认branching=32，iterations=11，centers_init = CENTERS_RANDOM, cb_index =0.2
FLANN_INDEX_COMPOSITE	随机kd树和层次k均值树来构建索引。默认trees =4，branching =32，iterations =11，centers_init = CENTERS_RANDOM,cb_index =0.2
FLANN_INDEX_KDTREE_SINGLE
FLANN_INDEX_HIERARCHICAL
FLANN_INDEX_LSH	multi-probe LSH方法
FLANN_INDEX_SAVED
FLANN_INDEX_AUTOTUNED
LINEAR
KDTREE
KMEANS
COMPOSITE
KDTREE_SINGLE
SAVED
AUTOTUNED	自动选取，以提供最好的性能

searchParams
SearchParams (int checks=32, float eps=0, bool sorted=true)

参数	描述
checks	默认32
eps	默认为0
sorted	默认True

2.匹配方法

FlannBasedMatcher对象继承了cv::DescriptorMatcher.

matches	=	cv.DescriptorMatcher.knnMatch(	queryDescriptors, trainDescriptors, k[, mask[, compactResult]]	)
matches	=	cv.DescriptorMatcher.knnMatch(	queryDescriptors, k[, masks[, compactResult]]	)

参数	描述
queryDescriptors	原图
trainDescriptors	搜索的图片
matches	匹配的结果
k	Count of best matches found per each query descriptor or less if a query descriptor has less than k possible matches in total.设置阈值，越高精度越高，匹配的数量越少
masks	Set of masks.
compactResult	Parameter used when the mask (or masks) is not empty.

matches

参数	描述
distance
imgIdx	训练图片的索引
queryIdx	query descriptor index
trainIdx	train descriptor index

参考文献：
https://www.jianshu.com/p/d70d9c8b2bec
https://docs.opencv.org/3.4.3/d5/d03/structcv_1_1flann_1_1SearchParams.html
https://docs.opencv.org/3.4.3/dc/d8c/namespacecvflann.html#a4e3e6c98d774ea77fd7f0045c9bc7817af67e191e4866496ff7db55c24277f4a4
https://docs.opencv.org/master/db/d39/classcv_1_1DescriptorMatcher.html#a38f18a66b3498fa5c960e29ed0056d44