openCV之单应性矩阵

import cv2
import numpy as np

# 打开图片
img1 = cv2.imread('./image/min_cat.png')
img2 = cv2.imread('./image/cat.png')

# 灰度化
gray1 = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)
gray2 = cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY)

# 创建特征检测器
sift = cv2.xfeatures2d.SIFT_create()

# 把关键点(特征点)和描述子一起检测出来.
kp1, des1 = sift.detectAndCompute(gray1,None)
kp2, des2 = sift.detectAndCompute(gray2,None)

# 创建特征匹配器
index_params = dict(algorithm = 1, trees = 5)
search_params = dict(checks = 50)
flann = cv2.FlannBasedMatcher(index_params,search_params)

# 对描述子进行特征匹配
matches = flann.knnMatch(des1,des2,k=2)
# print(matches)

goods = []
for (m,n) in matches:
    # 阈值一般设0.7-0.8之间
    if m.distance < 0.75*n.distance:
        goods.append(m)

# print(goods)
# 通过goods把特征点找到
# 因为计算单应性矩阵要求最少4个点
if len(goods) >=4:
    src_points = np.float32([kp1[m.queryIdx].pt for m in goods]).reshape(-1,1,2)
    dst_points = np.float32([kp2[m.trainIdx].pt for m in goods]).reshape(-1,1,2)

    # 根据匹配上的关键点去计算单应性矩阵
    H , _ = cv2.findHomography(src_points,dst_points,cv2.RANSAC,5)
    # 通过单应性矩阵，计算小图在大图中的对应位置
    h,w = img1.shape[:2]
    pts = np.float32([[0,0],[0,h-1],[w-1,h-1],[w-1,0]]).reshape(-1,1,2)
    # warpPerspective 是对图片进行透视变换的
    # cv2.warpPerspective()
    dst = cv2.perspectiveTransform(pts,H)
    # 在大图中，把dst画出来
    cv2.polylines(img2,[np.int32(dst)],True,(0,0,255),2)

else:
    print('not enough point number to compute homography matrix')
    exit()

# 画出匹配的特征点
ret = cv2.drawMatchesKnn(img1,kp1,img2,kp2,[goods],None)

cv2.imshow('ret',ret)
cv2.imshow('img2',img2)

cv2.waitKey(0)
cv2.destroyAllWindows()