1. #!/usr/bin/env python
  2. # encoding: utf-8
  4. import argparse
  5. import cv2
  6. import imutils
  8. # construct the argument parser and parse the arguments
  9. ap = argparse.ArgumentParser()
  10. ap.add_argument("-i", "--image", required=False, default='tetris_blocks.png', help="path to input image")
  11. args = vars(ap.parse_args())
  13. # load the input image (whose path was supplied via command line
  14. # argument) and display the image to our screen
  15. image = cv2.imread(args["image"])
  16. cv2.imshow("Image", image)
  17. cv2.waitKey(0)
  18. # convert the image to grayscale
  19. gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
  20. cv2.imshow("Gray", gray)
  21. cv2.waitKey(0)
  24. # applying edge detection we can find the outlines of objects in
  25. # images
  26. edged = cv2.Canny(gray, 30, 150)
  27. cv2.imshow("Edged", edged)
  28. cv2.waitKey(0)
  31. # threshold the image by setting all pixel values less than 225
  32. # to 255 (white; foreground) and all pixel values >= 225 to 255
  33. # (black; background), thereby segmenting the image
  34. thresh = cv2.threshold(gray, 225, 255, cv2.THRESH_BINARY_INV)[1]
  35. cv2.imshow("Thresh", thresh)
  36. cv2.waitKey(0)
  39. # a typical operation we may want to apply is to take our mask and
  40. # apply a bitwise AND to our input image, keeping only the masked
  41. # regions
  42. mask = thresh.copy()
  43. output = cv2.bitwise_and(image, image, mask=mask)
  44. cv2.imshow("Output", output)
  45. cv2.waitKey(0)
  48. # we apply erosions to reduce the size of foreground objects
  49. mask = thresh.copy()
  50. mask = cv2.erode(mask, None, iterations=5)
  51. cv2.imshow("Eroded", mask)
  52. cv2.waitKey(0)
  54. # similarly, dilations can increase the size of the ground objects
  55. mask = thresh.copy()
  56. mask = cv2.dilate(mask, None, iterations=5)
  57. cv2.imshow("Dilated", mask)
  58. cv2.waitKey(0)
  61. # find contours (i.e., outlines) of the foreground objects in the
  62. # thresholded image
  63. cnts = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
  64. cnts = imutils.grab_contours(cnts)
  65. output = image.copy()
  66. # loop over the contours
  67. for c in cnts:
  68.     # draw each contour on the output image with a 3px thick purple
  69.     # outline, then display the output contours one at a time
  70.     cv2.drawContours(output, [c], -1, (240, 0, 159), 3)
  71.     cv2.imshow("Contours", output)
  72.     cv2.waitKey(0)
  73.     pass
  75. # draw the total number of contours found in purple
  76. text = "I found {} objects!".format(len(cnts))
  77. cv2.putText(output, text, (10, 25),  cv2.FONT_HERSHEY_SIMPLEX, 0.7, (240, 0, 159), 2)
  78. cv2.imshow("Contours", output)
  79. cv2.waitKey(0)

image.png

image.png

ROI

Extracting “regions of interest” (ROIs) is an important skill for image processing.

Processing

Take note that drawing operations on images are performed in-place. Therefore at the beginning of each code block, we make a copy of the original image storing the copy

image.png

Canny

parameters to the cv2.Canny function:

  • img : The gray image.
  • minVal : A minimum threshold, in our case 30 .
  • maxVal : The maximum threshold which is 150 in our example.
  • aperture_size : The Sobel kernel size. By default this value is 3.

Thresholding

  1. # threshold the image by setting all pixel values less than 225
  2. # to 255 (white; foreground) and all pixel values >= 225 to 255
  3. # (black; background), thereby segmenting the image
  4. thresh = cv2.threshold(gray, 225, 255, cv2.THRESH_BINARY_INV)[1]
  5. cv2.imshow("Thresh", thresh)
  6. cv2.waitKey(0)
  • Grabbing all pixels in the gray image greater than 225 and setting them to 0 (black) which corresponds to the background of the image.
  • Setting pixel vales less than 225 to 255 (white) which corresponds to the foreground of the image.

Contours

Erosions and dilations

Erosions and dilations are typically used to reduce noise in binary images (a side effect of thresholding).

Masking and bitwise