How do I find the contours of a letter

Python
,
1

I used the findcontours method but it does not work as expected. What I want is the sketched version of the letter “v” from the original image.

This is the original image of letter “v”

image
image1200×900 1.85 KB

After using findcontours, codes attached below, it gave me an image that is not exactly what I want. I am a new user so can only upload one image. Pls see replies below.

gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
ret, binary = cv.threshold(gray, 0, 255, cv.THRESH_BINARY_INV | cv.THRESH_OTSU)
thinned = cv.ximgproc.thinning(binary)
contours, hireachy = cv.findContours(thinned, cv.RETR_EXTERNAL, cv.CHAIN_APPROX_SIMPLE)
contours = extractContours(contours)

def extractContours(contours):
    contours = contours[0].tolist()
    x = []
    y = []
    newContours = []
    for i in contours:
        newContours.append(i[0])
    for i in newContours:
        x.append(i[0])
        y.append(i[1])
    return np.asfortranarray([x, y])
2

The above codes output this image below in the variable “contours”
image

Apparently, this looks like a “Y” more. This is the first issue.
The second issue is that, this output image, or “contours” variable in the codes above, actually stores points in a “round trip” way. By that I mean “contours” has points in the way like " (1,1) (3,3) (5,5) (7,7) (9,9) (8,8) (6,6) (4,4) (2,2) (0,0).

3

To give another illustration of “round trip”, below I extract each segment from “contours”, and I got these images below. As you can see, contours store points of “v” twice. But there are no duplicated points. (The reason why the blue segments are a little bit off is because I use Bézier curve so it looks a bit off.)
So I need some help to get what I want, that is a sketched version of a letter, for which the points can be stored in a “one way” trip in stead of a “round way” trip. . I am new user to opencv. Thanks for help!
img058-047-001

4

img058-047-002

5

img058-047-003

6

img058-047-004

7

img058-047-005

8

img058-047-006

9

img058-047-007

10

img058-047-008