3H proficient in opencv (V) - perspective transformation

0、 preparation

Right click the new project , choice Python File, Create a new one Python file , Then at the beginning import cv2 Import cv2 library ,import numpy And rename to np.

import cv2import numpy as np

We also need to know where OpenCV in , The direction of the coordinate axis is x Shaft to the right ,y Axis down , The coordinate origin is in the upper left corner , For example, the following one is 640 Pixels , Wide for 480 Pixel image .OK, Let's begin with this section .

1、 Scene description

Given any image , Divide the area you need from it , And display it in the specified size . Moreover, the segmented image is parallel to the displayed window in the linear space . Like ： I want to get the following K, In the original picture K It's lying obliquely , After segmentation, it should be displayed standing vertically . So how to achieve this effect ？

Before achieving the above effect , We need to get some important information , For example, playing cards in the picture above K Four coordinate points of . We can use the drawing tool to open this picture , Then place the mouse over the four points of the divided area , You can get the coordinate information .

 Because the width and height axes of the image in the drawing are OpenCV It's the same , When I put the mouse over K At the top left corner , The bottom left corner of the drawing shows （734 Pixels ,185 Pixels ）, This is the coordinate of the point in the upper left corner .

2、 Perspective transformation

The implementation of the above requirements seems difficult , But with cv2 library , In fact, only 9 Line code .

1. First, let's set the width of the output picture width by 250, high height by 350.
2. Then read our pictures , This is the picture above .
3. Define the four point coordinates of the target area , Respectively ： Top left 、 The upper right 、 The lower left 、 The lower right . We've got... In the last step .
4. pst2 Is the position of the output picture in the picture box , Select tile , Fixed writing .
5. Use library functions warpPerspective Realize perspective transformation , If you are interested, you can have an in-depth understanding of the principle .
6. Finally, output and display two images .

width,height=250,350img=cv2.imread("Resources/poker.jpg")pst1=np.float32([[734,185],[1101,266],[621,679],[1030,789]])pst2=np.float32([[0,0],[width,0],[0,height],[width,height]])matrix=cv2.getPerspectiveTransform(pst1,pst2)imgOutput=cv2.warpPerspective(img,matrix,(width,height))cv2.imshow("Image",img)cv2.imshow("Output",imgOutput)cv2.waitKey(0)

Let's look at the effect , The picture on the left is our perspective transformation image , On the right is the original image ：

The content of section 5 , It's really interesting , There are many applications in life . A thousand eyes is better than a thousand hands , Go and knock it again ~