Practical Python and OpenCV: Case Studies by Adrian Rosebrock

Author:Adrian Rosebrock
Language: eng
Format: mobi
Published: 0101-01-01T00:00:00+00:00

Next up, Hank needs to perform some preprocessing on the digit images:

Listing 6.3: dataset.py

13 def deskew(image, width):

14 (h, w) = image.shape[:2]

15 moments = cv2.moments(image)

16

17 skew = moments["mu11"] / moments["mu02"]

18 M = np.float32([

19 [1, skew, -0.5 * w * skew],

20 [0, 1, 0]])

21 image = cv2.warpAffine(image, M, (w, h),

22 flags = cv2.WARP_INVERSE_MAP | cv2.INTER_LINEAR)

23

24 image = imutils.resize(image, width = width)

25

26 return image

Everybody has a different writing style. While most of us write digits that “lean” to the left, some “lean” to the right. Some of us write digits at varying angles. These varying angles can cause confusion for the machine learning models trying to learn the representation of various digits.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.