checkWeather.py

import cv2
import os
import numpy as np
import imutils
import logging
import time
from PIL import Image
from walkerArgs import parseArgs
import glob, os

log = logging.getLogger(__name__)

args = parseArgs()

weatherImages = {
    'weatherIcon_small_sunny.png' :1 ,
    'weatherIcon_small_rain.png' : 2,
    'weatherIcon_small_partlycloudy_day.png' : 3,
    'weatherIcon_small_cloudy.png' : 4,
    'weatherIcon_small_windy.png' : 5,
    'weatherIcon_small_snow.png' : 6,
    'weatherIcon_small_fog.png' : 7,
    'weatherIcon_small_clear.png' : 11,
    'weatherIcon_small_partlycloudy_night.png' : 13,
    'weatherIcon_small_extreme.png' : 16
}

def weather_image_matching(weather_icon_name, screenshot_name):

    weather_icon = cv2.imread(weather_icon_name,3)

    if (weather_icon is None):
        log.error('weather_image_matching: %s appears to be corrupted' % str(url_img_name))
        return 0

    screenshot_img = cv2.imread(screenshot_name,3)

    if (screenshot_img is None):
        log.error('weather_image_matching: %s appears to be corrupted' % str(screenshot_name))
        return 0
    height, width,_ = weather_icon.shape
    
    fort_img = imutils.resize(screenshot_img, width = int(screenshot_img.shape[1] * 2))
    height_f, width_f,_ = screenshot_img.shape
    screenshot_img = screenshot_img[0:height_f/7,0:width_f]


    resized = imutils.resize(weather_icon, width = int(weather_icon.shape[1] * 1))
    
    crop = cv2.Canny(resized, 100, 200)
        
    if crop.mean() == 255 or crop.mean() == 0:
        return 0

    (tH, tW) = crop.shape[:2]
    
    screenshot_img = cv2.blur(screenshot_img,(3,3))
    screenshot_img = cv2.Canny(screenshot_img, 50, 100)
  

    found = None
    for scale in np.linspace(0.2,1, 5)[::-1]:
        resized = imutils.resize(screenshot_img, width = int(screenshot_img.shape[1] * scale))
        r = screenshot_img.shape[1] / float(resized.shape[1])

        if resized.shape[0] < tH or resized.shape[1] < tW:
            break

        result = cv2.matchTemplate(resized, crop, cv2.TM_CCOEFF_NORMED)
        (_, maxVal, _, maxLoc) = cv2.minMaxLoc(result)
        
        (endX, endY) = (int((maxLoc[0] + tW) * r), int((maxLoc[1] + tH) * r))

        if endY > height_f/7 and endX < width_f/2:
            maxVal = 0

        if found is None or maxVal > found[0]:
	        found = (maxVal, maxLoc, r)

    return found[0]

def checkWeather(raidpic):
    foundweather = None
    
    for file in glob.glob(os.path.join('weather', 'weatherIcon_small_*.png')):
        filename = os.path.basename(file)
        find_weather= weather_image_matching(file,raidpic)

        if foundweather is None or find_weather > foundweather[0]:
            foundweather = find_weather, os.path.basename(file)
            
    if foundweather[0] > 0:
        weatherName = foundweather[1].split('.')
        log.info('The weather on the screenshot could be identified  (%s)' % str(weatherName[0].replace('_', ' ')))
        return True, weatherImages[os.path.basename(foundweather[1])]
        # True, WeatherID
        #send to database !
    else:
        log.error('The weather on the screenshot could not be identified')
        return False, False
        # nothing to do

    
if __name__ == '__main__':
   
    checkWeather(os.path.join(args.temp_path, 'screenshot.png'))