find_color.py

#!/usr/bin/env python
from __future__ import print_function, division

import cv2
import numpy as np
import rospy
from cv_bridge import CvBridge, CvBridgeError
from sensor_msgs.msg import Image
from std_msgs.msg import Float32

H_MIN, S_MIN, V_MIN = 0, 0, 0
H_MAX, S_MAX, V_MAX = 179, 255, 255
MIN_RADIUS_SIZE = 10

# Green Thresholds
GREEN_LOWER = [29, 86, 6]
GREEN_UPPER = [84, 255, 255]

# Dummy callback for trackbar initialization
def nothing(x):
    pass

class FindColor:
    def __init__(self, debug=False):
        # Converter of ros image to opencv image
        self.bridge = CvBridge()
        # Subscribe to the camera
        #TODO

        # Create the center_deviation topic where we will publish the error
        #TODO
        # Create the ball_radius topic where we will publish the radius of the ball
        self.ball_radius_pub = rospy.Publisher('ball_radius', Float32, queue_size = 10)
        # Dummy images
        self.cv_image = np.zeros((100, 100, 3), np.uint8)
        self.hsv = np.zeros((100, 100, 3), np.uint8)
        # Init Thresholds
        self.h_min, self.s_min, self.v_min = GREEN_LOWER
        self.h_max, self.s_max, self.v_max = GREEN_UPPER
        self.debug = debug
        if debug:
            self.image_name, self.mask_name = "RGB", "Mask"
            # Create Windows
            cv2.namedWindow(self.image_name)
            cv2.namedWindow(self.mask_name)
            cv2.waitKey(1000)
            # Creating track bar
            for min_max in ['min', 'max']:
                cv2.createTrackbar('h_{}'.format(min_max), self.mask_name, H_MIN, H_MAX, nothing)
                cv2.createTrackbar('s_{}'.format(min_max), self.mask_name, S_MIN, S_MAX, nothing)
                cv2.createTrackbar('v_{}'.format(min_max), self.mask_name, V_MIN, V_MAX, nothing)

            # Initial value, for green
            cv2.setTrackbarPos('h_min', self.mask_name, self.h_min)
            cv2.setTrackbarPos('s_min', self.mask_name, self.s_min)
            cv2.setTrackbarPos('v_min', self.mask_name, self.v_min)
            cv2.setTrackbarPos('h_max', self.mask_name, self.h_max)
            cv2.setTrackbarPos('s_max', self.mask_name, self.s_max)
            cv2.setTrackbarPos('v_max', self.mask_name, self.v_max)

    def callback(self, data):
        try:
            self.cv_image = self.bridge.imgmsg_to_cv2(data, "bgr8")
        except CvBridgeError as e:
            print("Error converting the image: {}".format(e))
            return
        if self.debug:
            # Get info from track bar and appy to result
            self.h_min = cv2.getTrackbarPos('h_min', self.mask_name)
            self.s_min = cv2.getTrackbarPos('s_min', self.mask_name)
            self.v_min = cv2.getTrackbarPos('v_min', self.mask_name)
            self.h_max = cv2.getTrackbarPos('h_max', self.mask_name)
            self.s_max = cv2.getTrackbarPos('s_max', self.mask_name)
            self.v_max = cv2.getTrackbarPos('v_max', self.mask_name)

        # Convert to HSV color space
        self.hsv = cv2.cvtColor(self.cv_image, cv2.COLOR_RGB2HSV)
        mask, center, radius = None, None, 0

        # If valid values update the Thresholds
        if self.h_min <= self.h_max and self.s_min <= self.s_max and self.v_min <= self.v_max:
            lower = np.array([self.h_min, self.s_min, self.v_min])
            upper = np.array([self.h_max, self.s_max, self.v_max])
            # Create the binary mask
            mask = cv2.inRange(self.hsv, lower, upper)
            # Erosion and dilation to remove noise
            kernel = np.ones((7, 7), np.uint8)
            mask = cv2.erode(mask, kernel, iterations=2)
            mask = cv2.dilate(mask, kernel, iterations=2)
            # Find contours in the mask
            contours = cv2.findContours(mask.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2]
            if len(contours) > 0:
                # Find largest contour and compute the minimum enclosing circle
                c = max(contours, key=cv2.contourArea)
                ((x, y), radius) = cv2.minEnclosingCircle(c)
                # Compute the centroid of the mask
                M = cv2.moments(c)
                center = (int(M["m10"] / M["m00"]), int(M["m01"] / M["m00"]))
            else:
                pass
                # print("No center found")
        else:
            print("Invalid values for Thresholds")

        # Create the masked image
        self.mask_image = cv2.bitwise_and(self.cv_image, self.cv_image, mask=mask)

        # Only proceed if the radius meets a minimum size
        if radius > MIN_RADIUS_SIZE and self.debug:
            # Draw the enclosing circle and center on the two images
            cv2.circle(self.mask_image, (int(x), int(y)), int(radius), (0, 255, 255), 2)
            cv2.circle(self.mask_image, center, 5, (0, 0, 255), -1)
            cv2.circle(self.cv_image, (int(x), int(y)), int(radius), (0, 255, 255), 2)
            cv2.circle(self.cv_image, center, 5, (0, 0, 255), -1)

        # Publish deviation from the middle of the image (= the error)
        if center is not None:
            # print("Center found: {}".format(center))
            # center_deviation_px = int(center[0] - image_width // 2)
            # Rescale the error to [-1, 1]
            image_width = self.cv_image.shape[1]
            max_deviation = image_width // 2
            center_deviation = (center[0] - max_deviation) / max_deviation
            # print("Center deviation : {}".format(center_deviation))

            # Publish the error to the corresponding topic
            #TODO

        # Publish the ball_radius
        if radius > MIN_RADIUS_SIZE:
            self.ball_radius_pub.publish(radius)


        if self.debug:
            # Show the images
            cv2.imshow(self.image_name, self.cv_image)
            cv2.imshow(self.masked, self.mask_image)
            # Wait 3 ms to show the images
            cv2.waitKey(3)


if __name__ == '__main__':

    print("Starting FindColor node")
    rospy.init_node('FindColor', anonymous=True)
    ic = FindColor()
    try:
        rospy.spin()
    except KeyboardInterrupt:
        print("Shutting down")
        cv2.destroyAllWindows()