crop.py

import cv2
import numpy as np
import sys

image = cv2.imread(sys.argv[1])
height, width, channels = image.shape
if width / height > 16 / 9:
  cv2.imwrite(sys.argv[2],image)
  exit()

imgray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)

# Set threshold
#th1 = cv2.adaptiveThreshold(imgray,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,1023,0)
_,th2 = cv2.threshold(imgray,8,255,cv2.THRESH_BINARY)
im2, contours, hierarchy = cv2.findContours(th2,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)

# Find with the largest rectangle
areas = [cv2.contourArea(contour) for contour in contours]
max_index = np.argmax(areas)
cnt = contours[max_index]
x,y,w,h = cv2.boundingRect(cnt)

# Ensure bounding rect should be at least 16:9 or taller
if w / h > 16 / 9:
  # increase top and bottom margin
  newHeight = w / 16 * 9
  y = y - (newHeight - h ) / 2
  h = newHeight


# Crop with the largest rectangle
crop = image[y:y+h,x:x+w]
cv2.imwrite(sys.argv[2],crop)

exit()
# Draw preview image
imagePreview = cv2.imread(sys.argv[1])
for contour in contours:
    #epsilon = 0.001 * cv2.arcLength(contour,True)
    #cv2.drawContours(imagePreview, [cv2.approxPolyDP(contour, epsilon, True)], 0, (0,255,0), 1)
    cv2.drawContours(imagePreview, [contour], 0, (0,255,0), 1)
cv2.imshow("adaptiveThreshold", th1)
cv2.imshow("globalThreshold", th2)
cv2.imshow("drawContours", imagePreview)
cv2.imshow("cropped", crop)
cv2.waitKey(0)
cv2.destroyAllWindows()