initial image manipulation commit

ocbot/pipeline/image/image_manipulation/image_edit.py

@@ -0,0 +1,134 @@
+import cv2
+import numpy as np
+
+
+def auto_canny(image, sigma=0.33):
+    # compute the median of the single channel pixel intensities
+    v = np.median(image)
+
+    # apply automatic Canny edge detection using the computed median
+    lower = int(max(0, (1.0 - sigma) * v))
+    upper = int(min(255, (1.0 + sigma) * v))
+    edged = cv2.Canny(image, lower, upper)
+
+    # return the edged image
+    return edged
+
+
+def find_white_pixels(canny_image):
+    return np.argwhere(canny_image == 255)
+    # reduce top, left, rigth and bottom size until we find the rectangular bound.
+
+
+def min_max_values(coord_list):
+    coord_dict = {'min_row': float('inf'),
+                  'max_row': float('-inf'),
+                  'min_col': float('inf'),
+                  'max_col': float('-inf'),
+                  }
+    for coord in coord_list:
+        if coord[0] < coord_dict['min_row']:
+            coord_dict['min_row'] = coord[0]
+
+        if coord[0] > coord_dict['max_row']:
+            coord_dict['max_row'] = coord[0]
+
+        if coord[1] < coord_dict['min_col']:
+            coord_dict['min_col'] = coord[1]
+        if coord[1] > coord_dict['max_col']:
+            coord_dict['max_col'] = coord[1]
+
+    return coord_dict
+
+
+def retain_aspect_scale(width, height, max_dim):
+    largest_dim = max(width, height)
+
+    ratio = max_dim/largest_dim
+    return ratio
+
+
+
+def change_background(cropped_img):
+
+    no_transparency =  cv2.cvtColor(cropped_img, cv2.COLOR_BGRA2BGR)
+    print(no_transparency.shape)
+    mask = cv2.cvtColor(no_transparency, cv2.COLOR_BGR2GRAY)
+    _, alpha = cv2.threshold(mask, 0, 255, cv2.THRESH_BINARY)
+
+    img_tuple = cv2.split(no_transparency)
+
+    b, g, r = img_tuple
+
+    #rgba = [b, g, r, alpha]
+    rgba = [b, g, r, alpha]
+    dst = cv2.merge(rgba, 3)
+    print('show')
+    cv2.imshow("image", dst)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
+    print('shown')
+    #dst= cv2.cvtColor(dst, cv2.COLOR_BGRA2BGR)
+
+    cv2.imwrite("test.png", dst)
+    return dst
+
+
+def pad_image(corrected_image, max_size):
+    old_size = corrected_image.shape[:2]
+
+
+    delta_w = max_size - old_size[1]
+    delta_h = max_size - old_size[0]
+
+
+    top, bottom = delta_h // 2, delta_h - (delta_h // 2)
+    left, right = delta_w // 2, delta_w - (delta_w // 2)
+
+    color = [255, 255, 255, 255]
+    new_im = cv2.copyMakeBorder(corrected_image, top, bottom, left, right, cv2.BORDER_CONSTANT,
+                                value=color)
+
+    cv2.imwrite("test.png", new_im)
+
+    img = cv2.imread('test.png')
+    blur = cv2.bilateralFilter(img, 9,75,75)
+
+    cv2.imwrite("test.png", new_im)
+
+    #return new_im
+    cv2.imshow("image", blur)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
+
+def main(img_input):
+    img = cv2.imread(img_input, 0)
+
+    edges = auto_canny(img)
+    min_bound = find_white_pixels(edges)
+
+    bound = min_max_values(min_bound)
+
+    img = cv2.imread(img_input, cv2.IMREAD_UNCHANGED)
+
+    crop_img = img[bound['min_row']:bound['max_row'], bound['min_col']: bound['max_col']]
+    scale_factor = retain_aspect_scale(bound['max_row'] - bound['min_row'], bound['max_col'] - bound['min_col'],
+                                       200 - 30)
+
+
+    scaled = cv2.resize(crop_img, None, fx=scale_factor, fy=scale_factor, interpolation=cv2.INTER_CUBIC)
+
+    corrected_image = change_background(scaled)
+    pad_image(corrected_image, 200)
+
+    # cv2.imshow("cropped", scaled)
+    #
+    # cv2.waitKey(0)
+
+
+if __name__ == '__main__':
+    imarr = ['1.png', '1.jpg', '2.png']
+
+    #imarr = ['ccamp.png', 'firehouse.png', 'guild.jpg']
+    for img in imarr:
+        main(img)