Related Articles

Related Articles

Contour Detection with Custom Seeds using Python – OpenCV
  • Last Updated : 01 Nov, 2020

This article discusses how we can dynamically detect contours by clicking on the Image and assign different colors to different parts of the image. Contours are a very useful tool for shape analysis and object detection and recognition.  This program uses OpenCV, Numpy, and Matplotlib libraries.  It also uses a built-in OpenCV Watershed algorithm to detect contours. 

Requirements

  • Python and OpenCV must be installed on the local machine.
  • Install Jupyter Notebook for easy debugging.
  • Here, Matplotlib’s colormap is used to get different colors. In the example given below, we will be using tab10. You can choose different colormaps. Refer to this site for different colors.

Explanation

  • Run the program.
  • Click on the image where you want to make contours.
  • Select a different color for a different part of the image by pressing numbers from zero to nine. (One number for one color)

Below is the Implementation.

Python3

filter_none

edit
close

play_arrow

link
brightness_4
code

# Import modules
import cv2
import numpy as np
import matplotlib.pyplot as plt
from matplotlib import cm
  
# Upload the image in the same directory and then run the code
# If image not found you may get an error
# Reading Image
road = cv2.imread('road_image.jpg')
  
# Making Copy of Image
road_copy = np.copy(road)
  
# Creating two black image of same size as original image
marker_image = np.zeros(road.shape[:2], dtype=np.int32)
segments = np.zeros(road.shape, dtype=np.uint8)
  
# Function to return tuple of colors
def create_rgb(i):
    x = np.array(cm.tab10(i))[:3]*255
    return tuple(x)
  
  
# Storing Colors
colors = []
  
# One color for each single digit
for i in range(10):
    colors.append(create_rgb(i))
  
# Global Variables
# Color Choices
# Number of markers
no_markers = 10
  
# Current markers
current_marker = 1
  
# Flag
marks_updated = False
  
# CALLBACK FUNCTION
def mouse_callback(event, x, y, flags, param):
    global marks_updated
  
    if event == cv2.EVENT_LBUTTONDOWN:
          
        # TRACKING FOR MARKERS
        cv2.circle(marker_image, (x, y), 5, (current_marker), -1)
          
        # DISPLAY ON USER IMAGE
        cv2.circle(road_copy, (x, y), 5, colors[current_marker], -1)
  
        marks_updated = True
  
  
# Naming the window and setting call back funtion to it
cv2.namedWindow('Road Image')
cv2.setMouseCallback('Road Image', mouse_callback)
  
while True:
    
    # Show the 2 windows
    cv2.imshow('WaterShed Segments', segments)
    cv2.imshow('Road Image', road_copy)
  
    # Close everything if Esc is pressed
    k = cv2.waitKey(1)
    if k == 27:
        break
  
    # Clear all colors and start over if 'c' is pressed
    elif k == ord('c'):
        road_copy = road.copy()
        marker_image = np.zeros(road.shape[0:2], dtype=np.int32)
        segments = np.zeros(road.shape, dtype=np.uint8)
  
    # If a number 0-9 is chosen index the color
    elif k > 0 and chr(k).isdigit():
          
        # chr converts to printable digit
        current_marker = int(chr(k))
  
    # If we clicked somewhere, call the watershed 
    # algorithm on our chosen markers
    if marks_updated:
        marker_image_copy = marker_image.copy()
        cv2.watershed(road, marker_image_copy)
        segments = np.zeros(road.shape, dtype=np.uint8)
  
        for color_ind in range(no_markers):
            
            # COLORING SEGMENTS
            segments[marker_image_copy == (color_ind)] = colors[color_ind]
  
        marks_updated = False
  
# Destroy all the windows at the end
cv2.destroyAllWindows()

chevron_right


Output:

This code will open two windows. One with the original image and one black. Clicking on the original image will create small circles on it and the contours will show on the black image. (Press numbers from 0-9 to change colors and produce a different contour plot.)



Original Image

Processed Image


Attention geek! Strengthen your foundations with the Python Programming Foundation Course and learn the basics.

To begin with, your interview preparations Enhance your Data Structures concepts with the Python DS Course.




My Personal Notes arrow_drop_up
Recommended Articles
Page :