Perspective transformation

Perspective transformation is also called projection transformation. The affine transformation we often talk about is a special case of perspective transformation. The purpose of perspective transformation is to transform objects that are straight lines in reality into straight lines in the picture. Perspective transformation can map a rectangle to an arbitrary quadrilateral.

This technology is used when the robot is driving automatically. Perspective transformation is done through the function:

dst = cv2. warpPerspective(src, M, dsize[,flag, [,borderMode[,borderValue]]])

dst : The output image after perspective transformation, dsize determines the actual size of the output.

src： Source Image

M：3X3 transformation matrix

dsize: Output image size.

flags： Interpolation method, the default is INTER_LINEAR (bilinear interpolation). When it is WARP_INVERSE_MAP, it means that M is an inverse transformation, which can achieve the inverse transformation from the target dst to src.

borderMode： The edge type. The default is BORDER_CONSTANT. When the value is BORDER_TRANSPARENT, the values in the target image are not changed, and these values correspond to the outliers in the original image.

borderValue： Boundary value, the default is 0. Like affine transformation, OpenCV still provides a function cv2.getPerspectiveTransform() to provide the transformation matrix above.

The function is as follows.

matAffine = cv2.getPerspectiveTransform(matSrc, matDst)

matSrc： The coordinates of the four vertices of the input image.

matDst： Output the coordinates of the four vertices of the image.

Code path:

x
~/dofbot_pro/dofbot_opencv/scripts/2.Transform/07_perspectivity.ipynb

​x
import cv2
import   numpy as np
import   matplotlib.pyplot as plt
img = cv2.imread('yahboom.jpg',1)

imgInfo =   img.shape
height =   imgInfo[0]
width =   imgInfo[1]
#src   4->dst 4 (Upper left corner, Lower left corner, Upper right corner, Lower right corner)
matSrc =   np.float32([[200,100],[200,400],[600,100],[width-1,height-1]])
matDst =   np.float32([[200,200],[200,300],[500,200],[500,400]])
#Combination

matAffine =   cv2.getPerspectiveTransform(matSrc,matDst)# mat 1 src 2 dst
dst =   cv2.warpPerspective(img,matAffine,(width,height))
img_bgr2rgb   = cv2.cvtColor(dst, cv2.COLOR_BGR2RGB)
plt.imshow(img_bgr2rgb)
    plt.show()