Sorting contours left to right in Python (OpenCV)

Question

I'm using Python and OpenCV to detect contours in my image. But when I run the following code to draw only a specific contour using the contour index, since the indices allocated are random, I get the wrong output.

So I found out the Centroids (in my case all the centroids lie in the same horizontal line). Is there any way I can sort the contour indices from left to right (starting from 0 to n) based on the x value of the centroid?

Could you please show the code for the same? Any help would be greatly appreciated!

contours, hierarchy = cv2.findContours(thresh,cv2.RETR_LIST,cv2.CHAIN_APPROX_SIMPLE)
i = 9         ## Contour Index <----

cv2.drawContours(img,contours[i:i+1],-1,(0,0,255),2)

centroids = []

for cnt in contours:        
    mom = cv2.moments(cnt)        
    (x,y) = int(mom['m10']/mom['m00']), int(mom['m01']/mom['m00'])         
    cv2.circle(org,(x,y),4,(255,255,255),-1)        
    centroids.append((x,y)

Answer 1

Implements de sort function of list in python like here.

In the implemented function you calculate the center and verify if the X position is gretter or smaller them the other. If gretter return 1, smaller -1 and equals 0.

def greater(a, b):
    momA = cv2.moments(a)        
    (xa,ya) = int(momA['m10']/momA['m00']), int(momA['m01']/momA['m00'])

    momB = cv2.moments(b)        
    (xb,yb) = int(momB['m10']/momB['m00']), int(momB['m01']/momB['m00'])
    if xa>xb 
        return 1

    if xa == xb
        return 0
    else
        return -1

For sure you can do better if you calculate de centers only once.

then just do

contours.sort(greater)