计算对象的数量 OpenCV - Python

Question

This is my first project in Python(3.5.1) and OpenCV (3), so I'm sorry for my mistakes. I've some pictures like these ones: https://s12.postimg.org/ox8gw5l8d/gado.jpg

I need to count how many white objects has on this image. I tried to ise SimpleBlobDetector but I didn't work as I was expecting.

# Standard imports
import cv2
import numpy as np;

# Read image
im = cv2.imread("C:/opencvTests/cattle.jpg", cv2.IMREAD_GRAYSCALE)

# Setup SimpleBlobDetector parameters.
params = cv2.SimpleBlobDetector_Params()

#filter by color
params.filterByColor = True
params.blobColor = 255

# Filter by Convexity
params.filterByConvexity = True
params.minConvexity = 0.87

# Filter by Inertia
params.filterByInertia = True
params.minInertiaRatio = 0.08

# Create a detector with the parameters
ver = (cv2.__version__).split('.')
if int(ver[0]) < 3 :
   detector = cv2.SimpleBlobDetector(params)
else : 
detector = cv2.SimpleBlobDetector_create(params)

# Detect blobs.
keypoints = detector.detect(im)

# Draw detected blobs as red circles.
# cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS ensures the size of the circle        corresponds to the size of blob
im_with_keypoints = cv2.drawKeypoints(im, keypoints, np.array([]),   (0,0,255),     cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

# Show keypoints
cv2.imwrite("C:/opencvTests/blobSave.jpg",im_with_keypoints)
print("Total of objects")
print(len(keypoints))

Any help would be really appreciate! Thanks in advance

Answer 1

Probably you will end up in wrong count if you continue working on this image. Perform some pre-processing operations such as morphological operations to remove noise and also to separate the object from each other. After this make use of "findcontours" an inbuilt opencv function. Then read the size of "findcontours" this will give the count of the objects.

Answer 2

I'm getting very close to the answer, I believe I just need to change some parameters according to the image. If someone needs something on this regard, this is my code:

# Standard imports
import cv2
import numpy as np;

# Read image
im = cv2.imread("C:/opencvTests/original.jpg", cv2.IMREAD_GRAYSCALE)

#Apply treshold
ret,im = cv2.threshold(im,240,255,cv2.THRESH_BINARY)

kernel = np.ones((6,6),np.uint8)
erosion = cv2.erode(im,kernel,iterations = 1)
opening = cv2.morphologyEx(im, cv2.MORPH_OPEN, kernel)
im = cv2.morphologyEx(opening, cv2.MORPH_CLOSE, kernel)


# Setup SimpleBlobDetector parameters.
params = cv2.SimpleBlobDetector_Params()

#filter by color
params.filterByColor = True
params.blobColor = 255

# Filter by Convexity
params.filterByConvexity = True
params.minConvexity = 0.87

# Filter by Inertia
params.filterByInertia = True
params.minInertiaRatio = 0.08

# Create a detector with the parameters
ver = (cv2.__version__).split('.')
if int(ver[0]) < 3 :
    detector = cv2.SimpleBlobDetector(params)
else : 
    detector = cv2.SimpleBlobDetector_create(params)

# Detect blobs.
keypoints = detector.detect(im)

# Draw detected blobs as red circles.
# cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS ensures the size of the circle        corresponds to the size of blob
im_with_keypoints = cv2.drawKeypoints(im, keypoints, np.array([]),   (0,0,255),     cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

# Show keypoints
cv2.imwrite("C:/opencvTests/keypoints.jpg",im_with_keypoints)
print("Total of objects")
print(len(keypoints))

Thank you very much!

Answer 3

I needed to implement the same logic. The task was to detect the number of houses from google map view.

Read the image in greyscale Apply threshold Apply morphological operations for noise reduction using cv2.findContour() function, the blobs were counted

Here's a part from the code which I wrote:

img = cv2.imread("images/map_snippet2.png", 0)

ret, thresh = cv2.threshold(img, 242, 255, cv2.THRESH_TOZERO_INV)   # every pixel below the threshold is turned white
ret, thresh_final = cv2.threshold(thresh, 240, 255, cv2.THRESH_TOZERO)  # every pixel below the threshold is turned black
# using the above 2 lines, we only took values in range [240, 242]

# applying morphological operaations to reduce noise
kernel = np.ones((3,3), np.uint8)
morph_img = cv2.erode(thresh_final, kernel, iterations=1)

# using findContours() function to count the white blobs. 
# using this function on morph_img is better than using it on canny_img

blob, hierarchy = cv2.findContours(morph_img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
blob_count = len(blob)

plt.imshow(morph_img, 'gray')   # this will count the number of white blobs in the images
plt.xlabel("Count = " + str(blob_count))
plt.ylabel("Detecting # of blobs")

plt.show()

Hope it helps