还未写完！！！

• 下面是两张需要拼接的图片
  
• 完整代码：

import cv2
import numpy as np
import sys


def cv_show(name, img):
    cv2.imshow(name, img)
    cv2.waitKey(0)


def detectAndDescribe(image):
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    descriptor = cv2.SIFT_create()

    (kps, des) = descriptor.detectAndCompute(gray, None)

    kps_float = np.float32([kp.pt for kp in kps])

    return (kps, kps_float, des)


''' 读取拼接图片 '''
imageA = cv2.imread('A.jpg')
cv_show('imageA', imageA)
imageB = cv2.imread('B.jpg')
cv_show('imageB', imageB)

''' 计算图片特征点及描述符 '''
(kpsA, kps_floatA, desA) = detectAndDescribe(imageA)
(kpsB, kps_floatB, desB) = detectAndDescribe(imageB)

''' 建立暴力匹配器BFMatcher,在匹配大型训练集时使用 FlannBasedMatcher 速度更快 '''
matcher = cv2.BFMatcher()

rawMatches = matcher.knnMatch(desB, desA, 2)
good = []
matches = []
for m in rawMatches:
    if len(m) == 2 and m[0].distance < 0.65 * m[1].distance:
        good.append(m)
        matches.append((m[0].trainIdx, m[0].queryIdx))
print(len(good))
print(matches)

vis = cv2.drawMatchesKnn(imageB, kpsB, imageA, kpsA, good, None, flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
cv_show("Keypoint Matches", vis)

""" 透视变换 """
if len(matches) > 4:    # 当筛选后的匹配对大于4时，计算视角变换矩阵
    # 获取匹配对的点坐标
    ptsA = np.float32([kps_floatA[i] for (i, _) in matches])
    ptsB = np.float32([kps_floatB[i] for (_, i) in matches])

    (H, mask) = cv2.findHomography(ptsB, ptsA, cv2.RANSAC, 10)
else:
    print('图片未找到4个以上的匹配点')
    sys.exit()

result = cv2.warpPerspective(imageB, H, (imageB.shape[1] + imageA.shape[1], imageB.shape[0]))
cv_show('resulB', result)

# 将图片A传入resultB图片最左端
result[0:imageA.shape[0], 0:imageA.shape[1]] = imageA
cv_show('result', result)

• 结果如下：