opencv_python FLANN近邻匹配笔记

FLANN邻近搜索

FLANN库全称是Fast Library for Approximate Nearest Neighbors，它是目前最完整的（近似）最近邻开源库。不但实现了一系列查找算法，还包含了一种自动选取最快算法的机制。
使用FLANN的搜索，整体来说分为两步，一是建立索引，二是搜索。

代码实现：

import cv2
import numpy as np
from matplotlib import pyplot as plt

def FLANN():
    queryImage = cv2.imread('img/left_01.png', 0)
    trainingImage = cv2.imread('img/right_01.png', 0)

    # 创建SIFT检测和计算
    sift = cv2.xfeatures2d.SIFT_create()
    kp1, des1 = sift.detectAndCompute(queryImage, None)
    kp2, des2 = sift.detectAndCompute(trainingImage, None)

    # FLANN匹配参数,定义FLANN匹配器，使用KNN算法实现匹配
    # 这里使用FLANN_INDEX_KDTREE，5kd-trees和50 checks迭代
    FLANN_INDEX_KDTREE = 1
    indexParams = dict(algorithm=1, trees=5)
    searchParams = dict(check=100)

    flann = cv2.FlannBasedMatcher(indexParams, searchParams)
    matches = flann.knnMatch(des1, des2, k=2)

    # 根据matches生成相同长度的matchesMask列表，列表元素为[0, 0]
    matchesMask = [[0, 0] for i in range(len(matches))]

    # 去除错误匹配
    # 比值检测认为第一个匹配和第二个匹配的比值小于一个给定的值（一般是0.5）
    for i, (m, n) in enumerate(matches):
        if m.distance < 0.5*n.distance:
            matchesMask[i] = [1, 0]
    # 将图像显示
    # mathColor是两图的匹配连接线，连接线与matchesMask相关
    # singlePointColor是勾画关键点
    drawParams = dict(matchColor=(0, 255, 0),
                      singlePointColor=(255, 0, 0),
                      matchesMask=matchesMask,
                      flags=0)
    resultImage = cv2.drawMatchesKnn(queryImage, kp1, trainingImage, kp2, matches,
                                     None, **drawParams)
    return resultImage

if __name__ == '__main__':
    resultImage = FLANN()
    plt.imshow(resultImage)
    plt.show()

单应性变换(Homography)

单应性是一个条件，该条件表明当两幅图像中的一副出现投影畸变时，它们还能彼此匹配。
一个平面到另一个平面的映射关系。
两张图分别有四个相对位置相同的点，Homography就是一个变换（3*3矩阵），将一张图中的点映射到另一张图中对应的点。


RANSAC
随机一致性采样RANSAC是一种鲁棒的模型拟合算法，能够从有外点的数据中拟合准确的模型。

代码实现：

import cv2
import numpy as np
from matplotlib import pyplot as plt

MIN_MATCH_COUNT = 10

queryImage = cv2.imread('img/box.png', 0)
trainingImage = cv2.imread('img/box_in_scene.png', 0)

sift = cv2.xfeatures2d.SIFT_create()
kp1, des1 = sift.detectAndCompute(queryImage, None)
kp2, des2 = sift.detectAndCompute(trainingImage, None)

FLANN_INDEX_KDTREE = 1
indexParams = dict(algorithm=1, trees=5)
searchParams = dict(check=100)

flann = cv2.FlannBasedMatcher(indexParams, searchParams)
matches = flann.knnMatch(des1, des2, k=2)

good = []
for m, n in matches:
    if m.distance < 0.5*n.distance:
        good.append(m)

if len(good) > MIN_MATCH_COUNT:
    src_pts = np.float32([kp1[m.queryIdx].pt for m in good]).reshape(-1, 1, 2)
    dst_pts = np.float32([kp2[m.trainIdx].pt for m in good]).reshape(-1, 1, 2)

    M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)
    matchesMask = mask.ravel().tolist()

    h, w = queryImage.shape
    pts = np.float32([[0, 0], [0, h-1], [w-1, h-1], [w-1, 0]]).reshape(-1, 1, 2)
    dst = cv2.perspectiveTransform(pts, M)

    trainingImage = cv2.polylines(trainingImage, [np.int32(dst)], True, 255, 3, cv2.LINE_AA)

else:
    print("Not enough matches are found %d/%d" % (len(good), MIN_MATCH_COUNT))
    matchesMask = None

draw_params = dict(matchColor=(0, 255, 0), singlePointColor=None, matchesMask=matchesMask, flags=2)

img3 = cv2.drawMatches(queryImage, kp1, trainingImage, kp2, good, None, **draw_params)

plt.imshow(img3, 'gray'), plt.show()

参考文章：

https://www.jianshu.com/p/57fc7068cfee