Update Arrow.py

Added Additional comments for better explanation.

Author: VyomGarg47

Scripts/Miscellaneous/Arrow_Angle_detector/Arrow.py

@@ -6,67 +6,74 @@
 cv2.namedWindow("Adjust") #Adjust help us to find the right threshold values for better accuracy using slider window.
 cv2.createTrackbar("min","Adjust",110,255, lambda x: None)
 
+def distance(x1,x2,y1,y2):
+	return (x1-x2)*(x1-x2) + (y1-y2)*(y1-y2)
+
+def exit():
+	cap.release()
+	cv2.destroyAllWindows()
+
 text = 'Left'
 while True:
 	ret ,frame = cap.read() #Reading each frame.
-	frame = cv2.resize(frame, (640,480), interpolation=cv2.INTER_AREA)
-	cropped = frame[100:400,100:500]
-	cropped = cv2.flip(cropped,1)
-	temp = cropped
+	frame = cv2.resize(frame, (640,480), interpolation=cv2.INTER_AREA) #Resizing
+	cropped = frame[100:400,100:500] #Cropping the frame
+	cropped = cv2.flip(cropped,1) #Flipping the frame vertically
+	temp = cropped #Temp is our original image
 	cropped = cv2.cvtColor(cropped, cv2.COLOR_BGR2GRAY)
-	mi = cv2.getTrackbarPos("min","Adjust")
+	mi = cv2.getTrackbarPos("min","Adjust") #Getting the values from the trackbar
 
-	_,threshold = cv2.threshold(cropped,mi,255,cv2.THRESH_BINARY)
-	kernal = np.zeros([10,10],np.uint8)
-	threshold = cv2.erode(threshold,kernal)
-	contours,_=cv2.findContours(threshold,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)
+	_,threshold = cv2.threshold(cropped,mi,255,cv2.THRESH_BINARY)#Converting image to binary (1's and 0's only)
+	kernal = np.zeros([10,10],np.uint8) #creating a square of 10X10 of zeroes
+	threshold = cv2.erode(threshold,kernal) #using this square to remove noise from the threshold
+	contours,_=cv2.findContours(threshold,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE) #Finding contours or areas with same color
 
-	for cnt in contours :
-		area = cv2.contourArea(cnt)
+	for cnt in contours : #iterating through each contour
+		area = cv2.contourArea(cnt) 
 		if area > 400 : #Selecting Contours only with area greater then 400
-			approx = cv2.approxPolyDP(cnt,0.01*cv2.arcLength(cnt,True),True)
+			approx = cv2.approxPolyDP(cnt,0.01*cv2.arcLength(cnt,True),True) #Reducing the number of edges/vertices or approximating the contour
 			approx_area = cv2.contourArea(approx)
 			if(len(approx)==7): #As an arrow has seven edges, hence we are selecting the polygon with 7 edges
-					n = approx.ravel()
+					n = approx.ravel() #Converts 2d array into 1d array
 					x1 = n[0] #First Co-ordinate always point to the topmost vertex
 					y1 = n[1] 
 					x2 = n[2]
 					y2 = n[3]
 					x3 = n[6]
 					y3 = n[7]
-					distance1 = (x1 - x2)*(x1 - x2) + (y1 - y2)*(y1 - y2) 
-					distance2 = (x1 - x3)*(x1 - x3) + (y1 - y3)*(y1 - y3)
+					distance1 = distance(x1,x2,y1,y2)
+					distance2 = distance(x1,x3,y1,y3)
 					ratio = distance1/distance2 #Ratio will help us to determine the orientation or angle of the arrow
 					if(2500<approx_area<25000 and (0.2<ratio<0.3 or ratio <0.1)):
-						cv2.drawContours(temp,[approx],0,(0,0,255),5)
+						cv2.drawContours(temp,[approx],0,(0,0,255),5) #Outlining the detected arrow with red color
 
-						x = approx.ravel()[0]
-						y = approx.ravel()[1]
+						x = approx.ravel()[0] #x co-ordinate of the top-most vertex (could be the tip or the end of the arrow)
+						y = approx.ravel()[1] #y co-ordinate of the top-most vertex (could be the tip or the end of the arrow)
 
-						if(0.2 < distance1/distance2 < 0.3): #Finding the tip of the arrow with the help of the ratios
+						if(0.2 < distance1/distance2 < 0.3): #Finding the tip of the arrow with the help of the ratios of the distances
 							cv2.putText(temp,"Arrow tip",(x,y),font,0.5,(0,0,255))
-							endx = (n[6]+n[8])/2
-							endy = (n[7]+n[9])/2
 							topx = x1
 							topy = y1
-							length = np.sqrt((topx-endx)*(topx-endx) + (topy-endy)*(topy-endy))
+							endx = (n[6]+n[8])/2
+							endy = (n[7]+n[9])/2
+							length = np.sqrt(distance(topx,endx,topy,endy))
 							y_v = endy-length
-							if((topx-endx)==0):
+							if((topx-endx)==0):#Denominator becomes zero
 								print(0)
 							else:
 								tan = (topy - y_v)/(topx - endx)
 								print(np.arctan(tan)*57.3*2) #Basic cirle property to print the angles ( between 90 and -90)
 
-						else :
+						else:
 							cv2.putText(temp,"Arrow end",(x,y),font,0.5,(0,0,255))
 							if(distance1/distance2 <0.1):
-								endx = (n[2]+n[0])/2
-								endy = (n[3]+n[1])/2
 								topx = n[8]
 								topy = n[9]
-								length = np.sqrt((topx-endx)*(topx-endx) + (topy-endy)*(topy-endy))
+								endx = (n[2]+n[0])/2
+								endy = (n[3]+n[1])/2
+								length = np.sqrt(distance(topx,endx,topy,endy))
 								y_v = endy - length
-								if((topx-endx)==0):
+								if((topx-endx)==0):#Denominator becomes zero
 									print(180)
 								else:
 									tan = (topy - y_v)/(topx - endx)
@@ -76,19 +83,16 @@
 								topy = n[7]
 								endx = (n[12]+n[0])/2
 								endy = (n[13]+n[1])/2
-								length = np.sqrt((topx-endx)*(topx-endx) + (topy-endy)*(topy-endy))
+								length = np.sqrt(distance(topx,endx,topy,endy))
 								y_v = endy - length
-								if((topx-endx)==0):
+								if((topx-endx)==0):#Denominator becomes zero
 									print(180)
 								else:
 									tan = (topy - y_v)/(topx - endx)
 									print(np.arctan(tan*57.3)*2)
 
-	cv2.imshow('temp',temp)
-	cv2.imshow('threshold',threshold)
-	#print(text)
-	if cv2.waitKey(1) & 0xFF == ord('q'):
+	cv2.imshow('temp',temp) #Displaying original image
+	cv2.imshow('threshold',threshold) #Displaying threshold image
+	if cv2.waitKey(1) & 0xFF == ord('q'): #Program will quite when q is pressed
 		break
-
-cap.release()
-cv2.destroyAllWindows()
+exit()