推荐|狗狗变形记：任选4点的投影变换warpPerspective OpenCV-Python案例

☞ ░ 老猿Python博文目录：https://blog.csdn.net/LaoYuanPython ░

要理解投影变换的原理最好能理解仿射变换的原理，关于仿射变换可参考老猿的仿射变换相关系列博文。

仿射变换博文传送门（带星号的为付费专栏文章）：

狗狗变形记：任选4点的投影变换warpPerspective案例

一、引言

在《http://iyenn.com/rec/324902.html openCV仿射变换：getAffineTransform的案例》我们介绍了任选三个点使用getAffineTransform获取仿射矩阵进行图像仿射变换，这节我们来选择四个点进行投影变换。

相关投影变换的概念、原理及OpenCV函数请参考《http://iyenn.com/rec/324894.html OpenCV-Python图像处理：透视变换概念、矩阵及实现案例详解》的介绍。

二、案例说明

2.1、输入图像

图像文件名：f:picdogandcat.JPG
在这里插入图片描述

2.2、功能说明

在图像上依次选择四个点，分别对应结果图像的左上、右上、左下、右下四个点，将这两组点作为参数使用getPerspectiveTransform获得投影变换（又称为透视变换）矩阵，然后调用warpPerspective进行投影变换。

三、示例代码

import cv2
import numpy as np

def OnMouseEvent( event, x, y, flags, param):
    global lbtDownPos
    global pos
    global pointList
    img = param
    ignoreEvent = [cv2.EVENT_MBUTTONDOWN, cv2.EVENT_MBUTTONUP, cv2.EVENT_MBUTTONDBLCLK, cv2.EVENT_MOUSEWHEEL,
                        cv2.EVENT_MOUSEHWHEEL,cv2.EVENT_MOUSEMOVE,cv2.EVENT_LBUTTONDBLCLK, cv2.EVENT_RBUTTONDBLCLK, cv2.EVENT_RBUTTONDOWN, cv2.EVENT_RBUTTONUP]  # 需要忽略的鼠标事件
    needRecordEvent = [ cv2.EVENT_LBUTTONDOWN, cv2.EVENT_LBUTTONUP]  # 需要记录当前信息的鼠标事件

    if event == cv2.EVENT_LBUTTONUP:
        pos = (x,y)
        print("OnMouseEvent EVENT_LBUTTONUP:",pos)
        n = len(pointList)
        if pos==lbtDownPos:
            n += 1
            if n <= 4:
                pointList.append(pos)
                cv2.putText(img, '.', (x-15, y+3), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=3, color=(255, 0, 0)) #坐标调整是因为点放大了占用了不止一个像素
                cv2.circle(img,(x , y),10,(0,0,255))
                cv2.putText(img, f'select point{n}:({x},{y})', (x + 20, y), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.3, color=(255, 0, 0))

        lbtDownPos = None

    elif event == cv2.EVENT_LBUTTONDOWN:
        lbtDownPos = (x,y)
        print("OnMouseEvent EVENT_LBUTTONDOWN:", lbtDownPos)

    else:lbtDownPos = None



def getPoint(imgfile):
    global pos
    global pointList

    pointList = []
    img = cv2.imread(imgfile)
    img = cv2.resize(img,None,fx=0.5,fy=0.5)
    cv2.putText(img, 'https://blog.csdn.net/LaoYuanPython', (100, 120), fontFace=cv2.FONT_HERSHEY_SIMPLEX,fontScale=0.5, color=(255, 0, 0))
    imgbak = np.array(img)
    rows,cols = img.shape[:2]
    winName = 'select 4 points'
    cv2.namedWindow(winName)
    cv2.setMouseCallback(winName, OnMouseEvent, img)
    print("请将要单独放大的部分从其左上角、右上角、左下角、右下角分别鼠标左键点击选择4个点，选择后在图像上有提示信息，选择完成后按ESC退出")


    while True:#通过鼠标左键点击选择4个点，分别代表要映射到左上角、右上角、左下角、右下角4个点
        cv2.imshow(winName, img)
        ch = cv2.waitKey(100)
        if ch == 27: break

    destPoint = [(0,0),(cols,0),(0,rows),(cols,rows)]
    if len(pointList)==4:
        pts1 = np.float32(pointList)
        pts2 = np.float32(destPoint)
        M = cv2.getPerspectiveTransform(pts1, pts2)
        dst = cv2.warpPerspective(imgbak, M, (cols, rows))
        cv2.imshow(winName, dst)
        ch = cv2.waitKey(0)
    else:
        print("没有选择足够的点")

getPoint(r'f:picdogandcat.JPG')
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
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67