Automatic Skew Correction Using Corner Detectors and Homography

The python file contains a skew corrector for rectangle-like objects. I have used the skew corrector on two types images

• Type 1: smooth with no interfering background
• Type 2: highly textured but more realistic image

Corner Detectors

I used the following two mechanisms of corner detection on image type 1 and image type 2 respectively

• Shi-tomashi corner detection using cv2.goodFeaturesToTrack()
• Contour detection and Douglas-Peucker algorithm using cv2.findContours() and cv2.approxPolyDP()

Homography and Image Transformation

• cv2.findHomography()
• cv2.warpPerspective()

Results

Figure 1

Figure 2

For detailed explanation refer to my Medium blog: https://blog.ekbana.com/skew-correction-using-corner-detectors-and-homography-fda345e42e65