Opencv-Python图像透视变换cv2.warpPerspective的示例

Opencv-Python图像透视变换cv2.warpPerspective

代码如下：

# -*- coding:utf-8 -*-
import cv2
import numpy as np
import sys
img = cv2.imread('test.jpg')
# cv2.imshow("original", img)
# 可选，扩展图像，保证内容不超出可视范围
img = cv2.copyMakeBorder(img, 200, 200, 200, 200, cv2.BORDER_CONSTANT, 0)
w, h = img.shape[0:2]
anglex = 0
angley = 30
anglez = 0 # 是旋转
fov = 42
r = 0
def rad(x):
 return x * np.pi / 180
def get_warpR():
 global anglex,angley,anglez,fov,w,h,r
 # 镜头与图像间的距离，21为半可视角，算z的距离是为了保证在此可视角度下恰好显示整幅图像
 z = np.sqrt(w ** 2 + h ** 2) / 2 / np.tan(rad(fov / 2))
 # 齐次变换矩阵
 rx = np.array([[1, 0, 0, 0],
         [0, np.cos(rad(anglex)), -np.sin(rad(anglex)), 0],
         [0, -np.sin(rad(anglex)), np.cos(rad(anglex)), 0, ],
         [0, 0, 0, 1]], np.float32)
 ry = np.array([[np.cos(rad(angley)), 0, np.sin(rad(angley)), 0],
         [0, 1, 0, 0],
         [-np.sin(rad(angley)), 0, np.cos(rad(angley)), 0, ],
         [0, 0, 0, 1]], np.float32)
 rz = np.array([[np.cos(rad(anglez)), np.sin(rad(anglez)), 0, 0],
         [-np.sin(rad(anglez)), np.cos(rad(anglez)), 0, 0],
         [0, 0, 1, 0],
         [0, 0, 0, 1]], np.float32)
 r = rx.dot(ry).dot(rz)
 # 四对点的生成
 pcenter = np.array([h / 2, w / 2, 0, 0], np.float32)
 p1 = np.array([0, 0, 0, 0], np.float32) - pcenter
 p2 = np.array([w, 0, 0, 0], np.float32) - pcenter
 p3 = np.array([0, h, 0, 0], np.float32) - pcenter
 p4 = np.array([w, h, 0, 0], np.float32) - pcenter
 dst1 = r.dot(p1)
 dst2 = r.dot(p2)
 dst3 = r.dot(p3)
 dst4 = r.dot(p4)
 list_dst = [dst1, dst2, dst3, dst4]
 org = np.array([[0, 0],
         [w, 0],
         [0, h],
         [w, h]], np.float32)
 dst = np.zeros((4, 2), np.float32)
 # 投影至成像平面
 for i in range(4):
   dst[i, 0] = list_dst[i][0] * z / (z - list_dst[i][2]) + pcenter[0]
   dst[i, 1] = list_dst[i][1] * z / (z - list_dst[i][2]) + pcenter[1]
 warpR = cv2.getPerspectiveTransform(org, dst)
 return warpR
def control():
 global anglex,angley,anglez,fov,r
 # 键盘控制
 if 27 == c: # Esc quit
   sys.exit()
 if c == ord('w'):
   anglex += 1
 if c == ord('s'):
   anglex -= 1
 if c == ord('a'):
   angley += 1
   print(angley)
   # dx=0
 if c == ord('d'):
   angley -= 1
 if c == ord('u'):
   anglez += 1
 if c == ord('p'):
   anglez -= 1
 if c == ord('t'):
   fov += 1
 if c == ord('r'):
   fov -= 1
 if c == ord(' '):
   anglex = angley = anglez = 0
 if c == ord('e'):
   print("======================================")
   print('Rotation Matrix:')
   print(r)
   print('angle alpha(anglex):')
   print(anglex)
   print('angle beta(angley):')
   print(angley)
   print('dz(anglez):')
   print(anglez)
while True:
 warpR = get_warpR()
 result = cv2.warpPerspective(img, warpR, (h, w))
 cv2.namedWindow('result',2)
 cv2.imshow("result", result)
 c = cv2.waitKey(30)
 control()
cv2.destroyAllWindows()

运行效果：

控制：

• s控制垂直方向上的形变

• a和d控制水平方向上的行变

• u和p控制角度旋转

• e 输出当前旋转矩阵参数

来源：https://blog.csdn.net/dcrmg/article/details/80273818