WIP #7: Basic Script for Removing Perceived Noise

"Perceived noise" used rather generously; this script compares the standard deviation of a pixel over a set of images and sets any that don't have a standard deviation of 0 to be white (grayscale value 255).

Author: Dankev55

vision/ImageCompareTest.py

@@ -0,0 +1,78 @@
+# This script compares a list of images and looks for any pixel whose standard
+# deviation over the set of pixels at that point is greater than 0, then sets it
+# to be white (255). It is for grayscale images, and runs slowly over large
+# images. 
+
+import cv2
+import numpy as np
+
+imgArr = []
+
+# NOTE: I did not upload the test images I used to the GitHub, because the
+# specific images used does not matter so long as they are grayscale.
+# Anyone can make images with paint (or just ask Daniel for the images).
+
+# Only ever have one set uncommented at once
+#First set of test images - small
+#img1 = cv2.imread ("TestImage1.png")
+#img2 = cv2.imread ("TestImage2.png")
+#img3 = cv2.imread ("TestImage3.png")
+#img4 = cv2.imread ("TestImage4.png")
+#img5 = cv2.imread ("TestImage5.png")
+
+#Second set of test images - medium
+img1 = cv2.imread ("TI6.png")
+img2 = cv2.imread ("TI7.png")
+img3 = cv2.imread ("TI8.png")
+img4 = cv2.imread ("TI9.png")
+img5 = cv2.imread ("TI10.png")
+
+#Third set of test images - large
+#img1 = cv2.imread ("TI11.png")
+#img2 = cv2.imread ("TI12.png")
+#img3 = cv2.imread ("TI13.png")
+#img4 = cv2.imread ("TI14.png")
+#img5 = cv2.imread ("TI15.png")
+
+#Just in case the images aren't actually grayscale
+#gray1 = cv2(img1, cv2.COLOR_BGR2GRAY)
+#gray2 = cv2(img2, cv2.COLOR_BGR2GRAY)
+#gray3 = cv2(img3, cv2.COLOR_BGR2GRAY)
+#gray4 = cv2(img4, cv2.COLOR_BGR2GRAY)
+#gray5 = cv2(img5, cv2.COLOR_BGR2GRAY)
+
+#imgArr.append(gray1)
+#imgArr.append(gray2)
+#imgArr.append(gray3)
+#imgArr.append(gray4)
+#imgArr.append(gray5)
+
+
+imgArr.append(img1)
+imgArr.append(img2)
+imgArr.append(img3)
+imgArr.append(img4)
+imgArr.append(img5)
+
+numImgs = 5                     # numImgs is the number of images used to test
+pixArr = [None] * numImgs
+
+leadImg = imgArr[0]                     #Uses img1 as a basis for iterating
+height, width = leadImg.shape[:2]
+for i in range(0, height):              #Iterates through entire image
+    for j in range(0,width):
+        for k in range (0,numImgs):
+            pixArr[k] = imgArr[k][i,j]  
+        if np.std(pixArr) > 0:
+            leadImg[i,j] = 255
+
+# To reduce clutter, you can comment out all the cv2.imshow lines besides
+# 'Adjusted.'
+cv2.imshow('Img1', imgArr[0])    # Possible concern? img1 is adjusted because
+cv2.imshow('Img2', imgArr[1])    # leadImg also accesses it
+cv2.imshow('Img3', imgArr[2])
+cv2.imshow('Img4', imgArr[3])
+cv2.imshow('Img5', imgArr[4])
+cv2.imshow('Adjusted', leadImg)
+cv2.waitKey(0)
+cv2.destroyAllWindows()