Question

所以我实现了this答案中描述的解决方案，它在某些情况下有效，而在其他情况下则失败。在我看来，在某些情况下，两个或两个以上的聚类正在求和以确定一个角落，而不是每个角落只有一个聚类。

这是一个图像（1），其簇位于拐角处：

Image（1），其中的簇正确地取平均值以获得中心：

但是，以下这些图像存在群集平均问题：

image（2），不平均群集：

image（2），具有平均聚类：仅在网格的左侧绘制中心

image（3），不包含平均群集：

image（3），平均簇：

当我尝试平均某些图像的群集时，我还会收到以下警告，然后出现程序崩溃：

C：\ Users \ ocheg \ OneDrive \ Desktop \ SpAM \ Solution1.py：52：RuntimeWarning：除以int_scalars中遇到的零（（x1-x2）（y3-y4）-（y1-y2）（x3-x4）） C：\ Users \ ocheg \ OneDrive \ Desktop \ SpAM \ Solution1.py：54：RuntimeWarning：无效> int_scalars中遇到的值（（x1-x2）（y3-y4）-（y1-y2）（x3-x4））

我的完整代码如下：

from PIL import Image
import numpy as np
import cv2
import glob

#import images from folder
images = [cv2.imread(file) for file in glob.glob("SpAMImages/*.jpg")]
for image in images:

#resize to make all images the same size, also makes display easier to work with
image = cv2.resize(image, (715,715))
#apply Gaussian blur
blur = cv2.GaussianBlur(image,(3,3),0)

#determines closing circle size. using 16 pixels
kernel = np.ones((7,7), np.uint8)

#converts image to grayscale
gray_img = cv2.cvtColor(blur, cv2.COLOR_BGR2GRAY)

#performs morphological closing
close_img = cv2.morphologyEx(gray_img, cv2.MORPH_CLOSE, kernel)

#applies Otsu thresholding
ret,threshold_img = cv2.threshold(close_img, 0, 255, cv2.THRESH_OTSU)

#closes holes with morphological closing
threshold_img = cv2.morphologyEx(threshold_img, cv2.MORPH_CLOSE, kernel)

#extract largest component from image.
components, output, stats, centroids = cv2.connectedComponentsWithStats(threshold_img, connectivity=4)
sizes = stats[:, -1]

max_label = 1
max_size = sizes[1]
for i in range(2, components):
    if sizes[i] > max_size:
        max_label = i
        max_size = sizes[i]
biggestComponent = np.zeros(output.shape)
biggestComponent[output == max_label] = 255
biggestComponent = biggestComponent - cv2.erode(biggestComponent, np.ones((5,5), np.uint8))
dilated = cv2.dilate(biggestComponent, np.ones((3,3), dtype=np.uint8))

#--------------------- obtaining corners using houghlines--------------------#
def find_intersection(line1, line2):
    # extract points
    x1, y1, x2, y2 = line1[0]
    x3, y3, x4, y4 = line2[0]

运行时警告在这里发生：

    # compute determinant
    Px = ((x1*y2 - y1*x2)*(x3-x4) - (x1-x2)*(x3*y4 - y3*x4))/  \
        ((x1-x2)*(y3-y4) - (y1-y2)*(x3-x4))
    Py = ((x1*y2 - y1*x2)*(y3-y4) - (y1-y2)*(x3*y4 - y3*x4))/  \
        ((x1-x2)*(y3-y4) - (y1-y2)*(x3-x4))
    return Px, Py

def segment_lines(lines, delta):
    h_lines = []
    v_lines = []
    for line in lines:
        for x1, y1, x2, y2 in line:
            if abs(x2-x1) < delta: # x-values are near; line is vertical
                v_lines.append(line)
            elif abs(y2-y1) < delta: # y-values are near; line is horizontal
                h_lines.append(line)
    return h_lines, v_lines

def cluster_points(points, nclusters):
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)
    _, _, centers = cv2.kmeans(points, nclusters, None, criteria, 10, cv2.KMEANS_PP_CENTERS)
    return centers

# run the Hough transform
lines = cv2.HoughLinesP(dilated.astype(np.uint8), rho=1, theta=np.pi/180, threshold=100, maxLineGap=50, minLineLength=150)

# segment the lines
delta = 180
h_lines, v_lines = segment_lines(lines, delta)

# draw the segmented lines
houghimg = image.copy()
for line in h_lines:
    for x1, y1, x2, y2 in line:
        color = [0,0,255] # color hoz lines red
        cv2.line(houghimg, (x1, y1), (x2, y2), color=color, thickness=1)
for line in v_lines:
    for x1, y1, x2, y2 in line:
        color = [255,0,0] # color vert lines blue
        cv2.line(houghimg, (x1, y1), (x2, y2), color=color, thickness=1)
# find the line intersection points
Px = []
Py = []
for h_line in h_lines:
    for v_line in v_lines:
        px, py = find_intersection(h_line, v_line)
        Px.append(px)
        Py.append(py)

# draw the intersection points
intersectsimg = image.copy()
for cx, cy in zip(Px, Py):
    cx = np.round(cx).astype(int)
    cy = np.round(cy).astype(int)
    color = np.random.randint(0,255,3).tolist() # random colors
    cv2.circle(intersectsimg, (cx, cy), radius=2, color=color, thickness=-1) # -1: filled circle
'''
# use clustering to find the centers of the data clusters
P = np.float32(np.column_stack((Px, Py)))
nclusters = 4
centers = cluster_points(P, nclusters)
#print(centers)

# draw the center of the clusters
for cx, cy in centers:
    cx = np.round(cx).astype(int)
    cy = np.round(cy).astype(int)
    cv2.circle(image, (cx, cy), radius=4, color=[0,0,255], thickness=-1) # -1: filled circle
'''
'''
biggestComponent = np.float32(biggestComponent)
dst = cv2.cornerHarris(biggestComponent,2,3,0.04)
dst = cv2.dilate(dst,None)
dst = cv2.dilate(dst,None)

#threshold
image[dst>0.01*dst.max()]=[0,0,255]
'''
'''
cv2.namedWindow('threshold image', cv2.WINDOW_NORMAL)
cv2.resizeWindow('threshold image', 400, 400)
cv2.namedWindow('biggest component', cv2.WINDOW_NORMAL)
cv2.resizeWindow('biggest component', 400, 400)
cv2.namedWindow('close_img', cv2.WINDOW_NORMAL)
cv2.resizeWindow('close_img', 400, 400)
cv2.namedWindow('hough', cv2.WINDOW_NORMAL)
cv2.resizeWindow('hough', 400, 400)
'''
cv2.namedWindow('intersection', cv2.WINDOW_NORMAL)
cv2.resizeWindow('intersection', 400, 400)
#cv2.imshow('threshold image', image)
#cv2.imshow('close_img', close_img)
#cv2.imshow('biggest component', biggestComponent)
#cv2.imshow('hough', houghimg)
cv2.imshow('intersection', intersectsimg)
cv2.waitKey(0)

群集平均和运行时警告

0 个答案: