poke_profile.py

# import the necessary packages
from PIL import Image
import pytesseract
import cv2
import numpy as np
from skimage.feature import match_template
from fuzzywuzzy import fuzz

orig_gray = None
resultImage = None
x = None
y = None
levelarea = None
def _parseLevel(gray, template):
	# find location of level
	global resultImage
	resultImage = match_template(gray, template)
	ij = np.unravel_index(np.argmax(resultImage), resultImage.shape)
	global x
	global y
	x, y = ij[::-1]
	# extract level from image
	height, width  = gray.shape
	global levelarea
	levelarea = gray[y-int(0.1*width):y+int(0.05*width),x-int(0.03*width):x+int(0.12*width)]
	level = pytesseract.image_to_string(levelarea)
	reqLevel = [int(s) for s in level.split() if s.isdigit()]
	return reqLevel

def getLevel(imagepath, username):
	result = [0,False]
	# load the example image and convert it to grayscale
	image = cv2.imread(imagepath)
	template_ios =  cv2.imread("template_ios.PNG")
	template_android = cv2.imread("template_android.PNG")
	if image is None:
		print("Error while reading file")
		return None
	gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
	global orig_gray
	orig_gray = gray
	gray_template_ios = cv2.cvtColor(template_ios, cv2.COLOR_RGB2GRAY)
	gray_template_android = cv2.cvtColor(template_android, cv2.COLOR_RGB2GRAY)

	# test for android template
	reqLevel = _parseLevel(gray, gray_template_android)
	if reqLevel == []:
		# if fail, test for iOS template
		reqLevel = _parseLevel(gray, gray_template_ios)
		if reqLevel == []:
			result[0] = None
		else:
			result[0] = reqLevel[0]
	else:
		result[0] = reqLevel[0]

	# apply thresholding to preprocess the image
	gray = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]

	# use tesseract to parse image to text
	text = pytesseract.image_to_string(gray)
	clearText = ''.join(e for e in text if e.isalnum())
	# fuzzy match username to text from image
	if (fuzz.partial_ratio(username, clearText) >= 80):
		# at least 80% match
		result[1] = True
	else:
		result[1] = False
	return result

# main if module used standalone
if __name__ == '__main__':
	import argparse
	import matplotlib.pyplot as plt

	ap = argparse.ArgumentParser()
	ap.add_argument("-i", "--image", required=True,
		help="path to input image to be OCR'd")
	ap.add_argument("-u", "--username", required=True,
		help="username to search in OCR output")
	ap.add_argument("-p", "--plot", type=str, default="false",
		help="true, to display plot, false else")
	args = vars(ap.parse_args())

	result = getLevel(args["image"],args["username"])
	if result is not None:
		print("Level Found: " + str(result[0]))
		if result[1]:
			print("Username found: " + args["username"])
		else:
			print("No Username found")

	# plott images
	if (args["plot"] == "true"):
		fig = plt.figure(figsize=(8, 3))
		ax1 = plt.subplot(1, 3, 1)
		ax2 = plt.subplot(1, 3, 3)
		ax3 = plt.subplot(1, 3, 2, sharex=ax2, sharey=ax2)
		ax1.imshow(orig_gray)
		ax2.imshow(levelarea)
		ax3.imshow(resultImage)
		ax3.set_axis_off()
		ax3.set_title('`match_template`\nresult')
		# highlight matched region
		ax3.autoscale(False)
		ax3.plot(x, y, 'o', markeredgecolor='r', markerfacecolor='none', markersize=10)

		plt.show()