【计算机视觉】基于SIFT算法的图像拼接小项目（精炼137行代码＆模块化＆拼接效果优良）

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


# 特征点提取函数
def extract_features(img1, img2):
    sift = cv.xfeatures2d.SIFT_create(nOctaveLayers=4)
    key1, describe1 = sift.detectAndCompute(img1, None)
    key2, describe2 = sift.detectAndCompute(img2, None)
    print(f"模板特征点数目：{len(key1)}")
    print(f"待匹配图片特征点数：{len(key2)}")


    img3 = img1.copy()  # 复制img1并命名为img3，在img3上绘制关键点
    img4 = img2.copy()  # 同上    img3 = cv.drawKeypoints(img3, key1, img3, )  # 绘制关键点
    img4 = cv.drawKeypoints(img4, key2, img4, )
    fig, axes = plt.subplots(1, 2)  # 创建一个1x2的子图布局，将两个图像放在不同的子图中
    axes[0].imshow(img3), axes[0].axis('off')  # 在第一个子图中显示img1
    axes[1].imshow(img4), axes[1].axis('off'), plt.show()  # 在第二个子图中显示img2

    return key1, describe1, key2, describe2,img3,img4



# 特征点匹配函数
def match_features(describe1, describe2, MIN=10):
    FLANN_INDEX_KDTREE = 0
    indexParams = dict(algorithm=FLANN_INDEX_KDTREE, trees=5)
    searchParams = dict(checks=50)
    flann = cv.FlannBasedMatcher(indexParams, searchParams)
    match = flann.knnMatch(describe1, describe2, k=2)
    print(f"总匹配对数：{len(match)}")

    good_matches = []
    for i, (m, n) in enumerate(match):
        if m.distance < 0.55 * n.distance:
            good_matches.append(m)
    print(f"正确匹配对数：{len(good_matches)}")

    return good_matches