In this image I am trying to detect horizontal lines. The code works well when image is not skewed. However, it is not working on such skewed images. I have tried this method to detect the right angle by histogram but many times is actually making it more skewed - python-opencv-skew-correction-for-ocr
Below is code to detect horizontal lines:
gray=cv2.cvtColor(img_final_bin,cv2.COLOR_BGR2GRAY)
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (100,1))
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
detected_lines = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=2)
cnts, hierarchy = cv2.findContours(detected_lines, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
boundingBoxes = [list(cv2.boundingRect(c)) for c in cnts]
Below is the code for skew correction, which is giving wrong results to me:
def correct_skew(image, delta=0.001, limit=3):
def determine_score(arr, angle):
data = inter.rotate(arr, angle, reshape=False, order=0)
histogram = np.sum(data, axis=1)
score = np.sum((histogram[1:] - histogram[:-1]) ** 2)
return histogram, score
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
print("thresh", thresh.shape)
thresh1 = thresh[0:500, 0:500]
print("thresh1", thresh1.shape)
scores = []
angles = np.arange(-limit, limit + delta, delta)
for i, angle in enumerate(angles):
histogram, score = determine_score(thresh1, angle)
scores.append(score)
# if i%100 == 0:
# print(score, angle, len(angles), i)
best_angle = angles[scores.index(max(scores))]
(h, w) = image.shape[:2]
center = (w // 2, h // 2)
M = cv2.getRotationMatrix2D(center, best_angle, 1.0)
rotated = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_CUBIC, \
borderMode=cv2.BORDER_REPLICATE)
return best_angle, rotated
Python Wand, which is based upon ImageMagick has a deskew function.
Input:
from wand.image import Image
from wand.display import display
with Image(filename='table.png') as img:
img.deskew(0.4*img.quantum_range)
img.save(filename='table_deskew.png')
display(img)
Result:
Related
I have used Projection Profile method on my binary images to get the deskew version. Everything is fine but the rotated image is having Black areas where the deskewing has been applied. How can I convert that area to white instead of black. Below is the code for Projection Profile.
def correct_skew(image, delta=1, limit=5):
"""
image : input
delta : sampling in the -limit,limit + delta range
limit : range of angles to explore
"""
# Function that returns the score of histogram for the given angle at which we check
def determine_score(arr, angle):
"""
arr : binarized image
angle : angle at which we calcuate the score
"""
data = inter.rotate(arr, angle, reshape=False, order=0)
histogram = np.sum(data, axis=1)
score = np.sum((histogram[1:] - histogram[:-1]) ** 2)
return histogram, score
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
scores = []
angles = np.arange(-limit, limit + delta, delta)
for angle in angles:
histogram, score = determine_score(thresh, angle)
scores.append(score)
best_angle = angles[scores.index(max(scores))]
(h, w) = image.shape[:2]
center = (w // 2, h // 2)
M = cv2.getRotationMatrix2D(center, best_angle, 1.0)
rotated = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_CUBIC)
return best_angle, rotated
This is the Image after Deskewing:
Original Binary Image:
The cv2.warpaffine documentation states that the function takes an optional argument namely borderValue. By default this value is (0, 0, 0), You can change this by calling your warpaffine routine as:
rotated = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_CUBIC, borderMode = cv2.BORDER_CONSTANT, borderValue=np.array([255, 255, 255]))
I'm working on a project about recognizing moroccan license plates which look like this image :
Moroccan License Plate
Please how can I use OpenCV to cut the license plate out and Tesseract to read the numbers and arabic letter in the middle.
I have looked into this research paper : https://www.researchgate.net/publication/323808469_Moroccan_License_Plate_recognition_using_a_hybrid_method_and_license_plate_features
I have installed OpenCV and Tesseract for python in Windows 10. When I run the tesseract on the text only part of the license plate using "fra" language I get 7714315l Bv. How can I separate the data?
Edit:
The arabic letters we use in Morocco are :
أ ب ت ج ح د هـ
The expected result is : 77143 د 6
The vertical lines are irrelevant, I have to use them to separate the image and read data separately.
Thanks in advance!
You can use HoughTransform since the two vertical lines are irrelevant, to crop the image:
import numpy as np
import cv2
image = cv2.imread("lines.jpg")
grayImage = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
dst = cv2.Canny(grayImage, 0, 150)
cv2.imwrite("canny.jpg", dst)
lines = cv2.HoughLinesP(dst, 1, np.pi / 180, 50, None, 60, 20)
lines_x = []
# Get height and width to constrain detected lines
height, width, channels = image.shape
for i in range(0, len(lines)):
l = lines[i][0]
# Check if the lines are vertical or not
angle = np.arctan2(l[3] - l[1], l[2] - l[0]) * 180.0 / np.pi
if (l[2] > width / 4) and (l[0] > width / 4) and (70 < angle < 100):
lines_x.append(l[2])
# To draw the detected lines
#cv2.line(image, (l[0], l[1]), (l[2], l[3]), (0, 0, 255), 3, cv2.LINE_AA)
#cv2.imwrite("lines_found.jpg", image)
# Sorting to get the line with the maximum x-coordinate for proper cropping
lines_x.sort(reverse=True)
crop_image = "cropped_lines"
for i in range(0, len(lines_x)):
if i == 0:
# Cropping to the end
img = image[0:height, lines_x[i]:width]
else:
# Cropping from the start
img = image[0:height, 0:lines_x[i]]
cv2.imwrite(crop_image + str(i) + ".jpg", img)
I am sure you know now how to get the middle part ;)
Hope it helps!
EDIT:
Using some morphological operations, you can also extract the characters individually:
import numpy as np
import cv2
image = cv2.imread("lines.jpg")
grayImage = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
dst = cv2.Canny(grayImage, 50, 100)
dst = cv2.morphologyEx(dst, cv2.MORPH_RECT, np.zeros((5,5), np.uint8),
iterations=1)
cv2.imwrite("canny.jpg", dst)
im2, contours, heirarchy = cv2.findContours(dst, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_NONE)
for i in range(0, len(contours)):
if cv2.contourArea(contours[i]) > 200:
x,y,w,h = cv2.boundingRect(contours[i])
# The w constrain to remove the vertical lines
if w > 10:
cv2.rectangle(image, (x, y), (x+w, y+h), (0, 0, 255), 1)
cv2.imwrite("contour.jpg", image)
Result:
This what I achieved by now...
The detection on second image was made by using the code found here: License plate detection with OpenCV and Python
Full code (which work from the third image an on) is this:
import cv2
import numpy as np
import tesserocr as tr
from PIL import Image
image = cv2.imread("cropped.png")
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
cv2.imshow('gray', image)
thresh = cv2.adaptiveThreshold(gray, 250, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 255, 1)
cv2.imshow('thresh', thresh)
kernel = np.ones((1, 1), np.uint8)
img_dilation = cv2.dilate(thresh, kernel, iterations=1)
im2, ctrs, hier = cv2.findContours(img_dilation.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
sorted_ctrs = sorted(ctrs, key=lambda ctr: cv2.boundingRect(ctr)[0])
clean_plate = 255 * np.ones_like(img_dilation)
for i, ctr in enumerate(sorted_ctrs):
x, y, w, h = cv2.boundingRect(ctr)
roi = img_dilation[y:y + h, x:x + w]
# these are very specific values made for this image only - it's not a factotum code
if h > 70 and w > 100:
rect = cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
clean_plate[y:y + h, x:x + w] = roi
cv2.imshow('ROI', rect)
cv2.imwrite('roi.png', roi)
img = cv2.imread("roi.png")
blur = cv2.medianBlur(img, 1)
cv2.imshow('4 - blur', blur)
pil_img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
api = tr.PyTessBaseAPI()
try:
api.SetImage(pil_img)
boxes = api.GetComponentImages(tr.RIL.TEXTLINE, True)
text = api.GetUTF8Text()
finally:
api.End()
# clean the string a bit
text = str(text).strip()
plate = ""
# 77143-1916 ---> NNNNN|symbol|N
for char in text:
firstSection = text[:5]
# the arabic symbol is easy because it's nearly impossible for the OCR to misunderstood the last 2 digit
# so we have that the symbol is always the third char from the end (right to left)
symbol = text[-3]
lastChar = text[-1]
plate = firstSection + "[" + symbol + "]" + lastChar
print(plate)
cv2.waitKey(0)
For arabic symbols you should install additional languages from TesseractOCR (and possibly use the version 4 of it).
Output: 77143[9]6
The number between brackets is the arabic symbol (undetected).
Hope I helped you.
I want to extract some rectangles at the top from a UML sequence diagram in jpg format by using OpenCV.
The algorithm I use finds way too many rectangles that are super small and not needed.
I think the mess up is somewhere in the beginning of the code where I apply canny edge detection but I am not sure.
I want to capture only the big rectangles from the top and center.
Thanks for any help.
import cv2
import numpy as np
import imutils
image = cv2.imread("./diagrams/sd2.jpg")
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
edges = cv2.Canny(gray, 90, 150, 3)
cnts = cv2.findContours(edges, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if imutils.is_cv2() else cnts[1]
cv2.drawContours(image, cnts, -1, (0, 255, 0), 1)
def detect(c):
shape = "unidentified"
peri = cv2.arcLength(c, True)
approx = cv2.approxPolyDP(c, 0.03 * peri, True)
if len(approx) == 4:
(x, y, w, h) = cv2.boundingRect(approx)
ar = w / float(h)
shape = "square" if ar >= 0.95 and ar <= 1.05 else "rectangle"
return shape
# loop over the contours
for c in cnts:
M = cv2.moments(c)
if M["m00"] != 0:
cX = int((M["m10"] / M["m00"]))
cY = int((M["m01"] / M["m00"]))
shape = detect(c)
c = c.astype("float")
c = c.astype("int")
if(shape == "rectangle"):
cv2.drawContours(image, [c], -1, (0, 255, 0), 2)
cv2.putText(image, shape, (cX, cY), cv2.FONT_HERSHEY_SIMPLEX,
0.5, (0, 0, 0), 2)
# show the output image
cv2.imshow("Image", image)
cv2.waitKey(0)
I am trying to detect edges from the products on a shelf using histogram projections. But I am stuck at 2 levels.
The challenges that I m facing are:
How to get the longest non shelf segment from the image i.e Detect the width of the widest product on the shelf from the available one.
How to achieve morphological reconstruction using custom markers.To eliminate
all small horizontal segments, I am generating 2 markers which can be seen in 'markers.png' (Attached). With them, I am calculating the minimum of the reconstruction outputs from both the markers.
Need assistance on this.
Thanks a lot
Below is my python code for the same.
Below is my python code
********************************************************************************
import numpy as np
import cv2 as cv
from matplotlib import pyplot as plt
import math
# Read the input image
img = cv.imread('C:\\Users\\672059\\Desktop\\p2.png')
# Converting from BGR to RGB. Default is BGR.
# img_rgb = cv.cvtColor(img, cv.COLOR_BGR2RGB)
# Resize the image to 150,150
img_resize = cv.resize(img, (150, 150))
# Get the dimensions of the image
img_h, img_w, img_c = img_resize.shape
# Split the image on channels
red = img[:, :, 0]
green = img[:, :, 1]
blue = img[:, :, 2]
# Defining a vse for erosion
vse = np.ones((img_h, img_w), dtype=np.uint8)
# Morphological Erosion for red channel
red_erode = cv.erode(red, vse, iterations=1)
grad_red = cv.subtract(red, red_erode)
# Morphological Erosion for green channel
green_erode = cv.erode(green, vse, iterations=1)
grad_green = cv.subtract(green, green_erode)
# Morphological Erosion for blue channel
blue_erode = cv.erode(blue, vse, iterations=1)
grad_blue = cv.subtract(blue, blue_erode)
# Stacking the individual channels into one processed image
grad = [grad_red, grad_green, grad_blue]
retrieved_img = np.stack(grad, axis=-1)
retrieved_img = retrieved_img.astype(np.uint8)
retrieved_img_gray = cv.cvtColor(retrieved_img, cv.COLOR_RGB2GRAY)
plt.title('Figure 1')
plt.imshow(cv.bitwise_not(retrieved_img_gray), cmap=plt.get_cmap('gray'))
plt.show()
# Hough Transform of the image to get the longest non shelf boundary from the image!
edges = cv.Canny(retrieved_img_gray, 127, 255)
minLineLength = img_w
maxLineGap = 10
lines = cv.HoughLinesP(edges, 1, np.pi/180, 127, minLineLength=1, maxLineGap=1)
temp = img.copy()
l = []
for x in range(0, len(lines)):
for x1, y1, x2, y2 in lines[x]:
cv.line(temp, (x1, y1), (x2, y2), (0, 255, 0), 2)
d = math.sqrt((x2-x1)**2 + (y2-y1)**2)
l.append(d)
# Defining a hse for erosion
hse = np.ones((1, 7), dtype=np.uint8)
opening = cv.morphologyEx(retrieved_img_gray, cv.MORPH_OPEN, hse)
plt.title('Figure 2')
plt.subplot(1, 2, 1), plt.imshow(img)
plt.subplot(1, 2, 2), plt.imshow(cv.bitwise_not(opening), 'gray')
plt.show()
# Dilation with disk shaped structuring element
horizontal_size = 7
horizontalstructure = cv.getStructuringElement(cv.MORPH_ELLIPSE, (horizontal_size, 1))
dilation = cv.dilate(opening, horizontalstructure)
plt.title('Figure 3')
plt.imshow(cv.bitwise_not(dilation), 'gray')
plt.show()
# Doing canny edge on dilated image
edge = cv.Canny(dilation, 127, 255)
plt.title('Figure 4')
plt.imshow(edges, cmap='gray')
plt.show()
h_projection = edge.sum(axis=1)
print(h_projection)
plt.title('Projection')
plt.plot(h_projection)
plt.show()
listing = []
for i in range(1, len(h_projection)-1):
if h_projection[i-1] == 0 and h_projection[i] == 0:
listing.append(dilation[i])
listing.append(dilation[i-1])
a = np.array([np.array(b) for b in l])
h = len(l)
_, contours, _ = cv.findContours(a, cv.RETR_EXTERNAL, cv.CHAIN_APPROX_SIMPLE)
x, y, w, h = cv.boundingRect(contours[0])
y = y + i - h
cv.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)
l.clear()
plt.imshow(img)
plt.show()
# Generating a mask
black_bg = np.ones([img_h, img_w], dtype=np.uint8)
# Clone the black bgd image
left = black_bg.copy()
right = black_bg.copy()
# Taking 10% of the image width
ten = int(0.1 * img_w)
left[:, 0:ten+1] = 0
right[:, img_w-ten:img_w+1] = 0
plt.title('Figure 4')
plt.subplot(121), plt.imshow(left, 'gray')
plt.subplot(122), plt.imshow(right, 'gray')
plt.show()
# Marker = left and right. Mask = dilation
mask = dilation
marker_left = left
marker_right = right
********************************************************************************
markers.png link: https://i.stack.imgur.com/45WJ6.png
********************************************************************************
Based on you input image, I would :
take a picture of an empty fridge
then compare the current image with the empty one.
play with morphological operations
get connected components > size N
If you can't take a empty fridge image:
segment the shelves (threshold white parts)
undo do the rotation of the image by using image moments of the shelves
for each shelve:
Threshold on saturation
Do a vertical projection
Count maxima.
Tresholded:
Erode-dilate:
Connected componens (width > 10 * height + > minsize):
And you have shelves.
Now take the average Y form each shelf and cut the original image in pieces:
Dither to 8 colors:
and threshold:
Connected components (h>1.5*w, minsize... this is hard here, I played with it :)
I have the above number plate image with me. My goal is to segment each character individually and pass into my neural network. I have tried to find countours and use bounding rectangles to segment these characters using the following code:
img = cv2.imread('download.jpeg')
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray_img = cv2.GaussianBlur(gray_img, (5,5), 0)
ret, im_th = cv2.threshold(gray_img, 90, 255, cv2.THRESH_BINARY_INV)
im_th = cv2.adaptiveThreshold(gray_img, 255,
cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY,75, 10)
im_th = cv2.bitwise_not(im_th)
ctrs, hier = cv2.findContours(im_th.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
cv2.drawContours(img, ctrs, -1, (0,255,0), 3)
rects = [cv2.boundingRect(ctr) for ctr in ctrs]
print len(rects)
for rect in rects:
cv2.rectangle(img,(rect[0], rect[1]), (rect[0] + rect[2], rect[1] + rect[3]), (0,255,0),3)
length = int(rect[3] * 1.6)
pt1 = int(rect[1] + rect[3] // 2 - length // 2)
pt2 = int(rect[0] + rect[2] // 2 - length // 2)
roi = img[pt1:pt1+length, pt2:pt2+length]
The above code creates regions that include bounding rectangles other than the characters. Although I could manually filter out these regions, it would vary from image to image. How would I go about this if I need to only extract the regions with characters?
#read image
img = cv2.imread('input_image.png')
#grayscale
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
cv2.waitKey(0)
#binarize
ret,thresh = cv2.threshold(gray,127,255,cv2.THRESH_BINARY_INV)
cv2.waitKey(0)
#find contours
im2,ctrs, hier = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
#sort contours
sorted_ctrs = sorted(ctrs, key=lambda ctr: cv2.boundingRect(ctr)[0])
for i, ctr in enumerate(sorted_ctrs):
# Get bounding box
x, y, w, h = cv2.boundingRect(ctr)
# Getting ROI
roi = img[y:y+h, x:x+w]
# show ROI
#cv2.imwrite('roi_imgs.png', roi)
cv2.imshow('charachter'+str(i), roi)
cv2.rectangle(img,(x,y),( x + w, y + h ),(90,0,255),2)
cv2.waitKey(0)
cv2.imshow('marked areas',img)
cv2.waitKey(0)