opencv定位图片中的图案？

使用opencv模板匹配方式。
这种方式针对从原始图片中直接扣小图情况比较好。

import cv2 as cv2

def find_positions(image_path, small_image_path):
    # 读取大图和小图
    large_image = cv2.imread(image_path)
    small_image = cv2.imread(small_image_path)

    # 小图规格
    small_image_h, small_image_w = small_image.shape[:2]

    # 对比大图与小图
    # 匹配模板
    res = cv2.matchTemplate(large_image, small_image, cv2.TM_CCOEFF_NORMED)
    min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
    # 计算矩形左边
    top_left = max_loc

    # 适用于小图长度>高度情况
    # bottom_right = (max_loc[0] + small_image_h, max_loc[1] + small_image_w)

    # 适用于小图长度<高度情况
    bottom_right = (max_loc[0] + small_image_w, max_loc[1] + small_image_h)
    # 画矩形
    cv2.rectangle(large_image, top_left, bottom_right, (0, 0, 255), 3)

    # 展示结果
    cv2.imshow('large_image', large_image)
    cv2.waitKey(0)
    pass


image_path = "D:/Desktop/big.png"
small_image_path = "D:/Desktop/small.png"
find_positions(image_path, small_image_path)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35

大图：

小图：

最后结果：