# PIL is a great python library for doing everything related to images # check out the other PIL and Image examples: # Find and Label objects in Images l # Find and outline the sun # Replace or remove colors from an image # Find the average RGB color for and image # Determine an image's type (regardless of extension) # here is the zipped full source for this example from PIL import Image import time import sys def compareTwoPics(picture_1, picture_2, step=20): step = int(step) if step < 1: step = 1 percent_similar = 0.0 percent_red = 0.0 percent_green = 0.0 percent_blue = 0.0 percent_totalcolor = 0.0 total_check_points = 0.0 total_not_match = 0.0 percent_similar = 0.0 pic1_total_red = 0 pic1_total_green = 0 pic1_total_blue = 0 pic1_total_color = 0 pic2_total_red = 0 pic2_total_green = 0 pic2_total_blue = 0 pic2_total_color = 0 #print picture_1 #print picture_2 try: pic1 = Image.open(picture_1) pic2 = Image.open(picture_2) except: # tried to open an unsupported format!! #print str(picture_1) + " is not a supported image format" #print str(picture_2) + " is not a supported image format" return -1.0, 0.0, 0.0, 0.0 pic1_width = pic1.size[0] pic1_height = pic1.size[1] pic2_width = pic2.size[0] pic2_height = pic2.size[1] if pic1_width != pic2_width: print "\n widths must be the same \n" sys.exit() if pic1_height != pic2_height: print "\n widths must be the same \n" sys.exit() for x in range(0, pic1_width): for y in range(0, pic1_height): if x % step == 0: pic1_color = pic1.getpixel((x,y)) pic2_color = pic2.getpixel((x,y)) total_check_points += 3 if pic1_color[0] != pic2_color[0]: total_not_match += 1 if pic1_color[1] != pic2_color[1]: total_not_match += 1 if pic1_color[2] != pic2_color[2]: total_not_match += 1 pic1_total_red += pic1_color[0] pic1_total_green += pic1_color[1] pic1_total_blue += pic1_color[2] pic2_total_red += pic2_color[0] pic2_total_green += pic2_color[1] pic2_total_blue += pic2_color[2] #print " pic1 count: "+ str(pic1_total_red) #print " pic2 count: "+ str(pic2_total_red) pic1_total_color = pic1_total_red + pic1_total_green + pic1_total_blue pic2_total_color = pic2_total_red + pic2_total_green + pic2_total_blue percent_similar = 1 - (total_not_match / total_check_points) percent_red = abs(float(pic2_total_red) / float(pic1_total_red)) percent_green = abs(float(pic2_total_green) / float(pic1_total_green)) percent_blue = abs(float(pic2_total_blue) / float(pic1_total_blue)) percent_totalcolor = abs(float(pic2_total_color) / float(pic1_total_color)) print '----' print "total % comparible red: " + str(percent_red) print "total % comparible green:" + str(percent_green) print "total % comparible blue: " + str(percent_blue) print "total % comparible: " + str(percent_totalcolor) print "total pic1: " + str(pic1_total_color) print "total pic2 " + str(pic2_total_color) return percent_similar, percent_totalcolor, pic1_total_color, pic2_total_color if __name__ == '__main__': try: import psyco psyco.full() except ImportError: print "...installing psyco would provide additional performance" thedetail = 20 if len( sys.argv ) < 3: print "\n\n" print "Usage: " + str(sys.argv[0]) + " <photo1.png> <photo2.png> [detail]" print "" print " - photo1.png and photo2.png are the photos you are comparing" print " - detail is just a number. The higher the number the faster" print " and less detailed the comparison will be. Default detail is 20" print "" print "example:" print " # Highest quality comparison, slowest turn around" print " " + str(sys.argv[0]) + " photo.png photo2.png 1" print " " print " # Low quality comparison, high turn around" print " " + str(sys.argv[0]) + " photo.png photo2.png 100" sys.exit() else: try: thedetail = sys.argv[3] except: thedetail = 20 tt = time.time() print str(compareTwoPics(sys.argv[1], sys.argv[2], step=thedetail)) print "execution seconds: " + str(time.time() - tt)
A python example based blog that shows how to accomplish python goals and how to correct python errors.
Showing posts with label psyco. Show all posts
Showing posts with label psyco. Show all posts
Wednesday, July 15, 2009
Python - compare two images
Subscribe to:
Posts (Atom)