CODEWITHSUNDEEP
pythonopencvimage-processingcomputer-visioncontour
Subham Tiwari
Subham Tiwari

Reputation: 460

Hand contour not precise

I wrote an OpenCV program to extract the hand out of the image precisely. But is not able to get it out correctly. Below is the code and the output and the sample image which I used to test it.

import numpy as np
import cv2
  
# Reading image
font = cv2.FONT_HERSHEY_COMPLEX
img2 = cv2.imread('1.bmp', cv2.IMREAD_COLOR)
  
# Reading same image in another 
# variable and converting to gray scale.
img = cv2.imread('1.bmp', cv2.IMREAD_GRAYSCALE)
  
# Converting image to a binary image
# ( black and white only image).
_, threshold = cv2.threshold(img, 110, 255, cv2.THRESH_BINARY)
  
# Detecting contours in image.
contours, _= cv2.findContours(threshold, cv2.RETR_TREE,
                               cv2.CHAIN_APPROX_SIMPLE)
contours1 = max(contours, key=cv2.contourArea)  
# Going through every contours found in the image.
  
approx = cv2.approxPolyDP(contours1, 0.009 * cv2.arcLength(contours1, True), True)

# draws boundary of contours.
cv2.drawContours(img2, [approx], 0, (0, 0, 255), 5) 

cv2.imshow('image2', img2) 
  
# Exiting the window if 'q' is pressed on the keyboard.
if cv2.waitKey(0) & 0xFF == ord('q'): 
    cv2.destroyAllWindows()

Input image -

Input Image

Output Image

Upvotes: 1

Views: 164

Answers (1)

Red
Red

Reputation: 27577

One of the reasons your contour is not precise is the obvious; the line where you approximated the contour. But you have also mentioned (in a comment) that lowering the approximation didn't solve the problem.

This is because you didn't blur the thresholded image, which resulted in the jagged edges. Here is an example where the thresholded image is blurred before the contour detection:

The code:

import cv2
import numpy as np

def process(img):
    img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    return cv2.threshold(img_gray, 111, 255, cv2.THRESH_BINARY)[1]

def draw_contours(img):
    contours, _ = cv2.findContours(process(img), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
    cnt = max(contours, key=cv2.contourArea)
    cv2.drawContours(img, [cnt], -1, (0, 0, 255), 2)

img = cv2.imread("image.png")
draw_contours(img)
cv2.imshow("result", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Input image:

enter image description here

Output image:

enter image description here

Still, the contour isn't very precise. This is where the Canny edge detector comes into play:

import cv2
import numpy as np

def process(img):
    img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    _, thresh = cv2.threshold(img_gray, 111, 255, cv2.THRESH_BINARY)
    img_blur = cv2.GaussianBlur(thresh, (5, 5), 4)
    img_canny = cv2.Canny(img_blur, 0, 0)
    img_dilate = cv2.dilate(img_canny, None, iterations=1)
    return cv2.erode(img_dilate, None, iterations=0)

def draw_contours(img):
    contours, _ = cv2.findContours(process(img), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
    cnt = max(contours, key=cv2.contourArea)
    cv2.drawContours(img, [cnt], -1, (0, 0, 255), 2)

img = cv2.imread("image.png")
draw_contours(img)
cv2.imshow("result", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

enter image description here

Upvotes: 1

Related Questions