How do I improve the number detection for blueprints (OCR) - python

I have a number of blueprints where I would like to detect the numbers on the blueprint such that I can turn them into proper models.
for example I have the following image and would like all the numbers on this image so I ran the following code:
import pytesseract
from pytesseract import Output
import cv2
import numpy as np
img = cv2.imread('vdb7C.jpg')
custom_config = r' (--oem 2 --psm 10'
d = pytesseract.image_to_data(img,config=custom_config,lang='eng', output_type=Output.DICT)
n_boxes = len(d['level'])
for i in range(n_boxes):
text=d["text"][i]
print(text+ str(str.isdigit(text)))
if str.isdigit(text):
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.imwrite("output.jpg" , img)
This gave me the following result: . As you can see it does correctly identify a number of numbers on the blueprint, however it misses quite a few others and falsely detect a few that aren't really there. I care more about getting all the numbers than a few false positives but would still like to keep those to a minimum so any suggestions there?
I have already tried thinning operations, re-scaling the images, rotating the images and smoothing the images but all of those don't appear to make much difference, extreme rescaling (*0.1 or *10) does change a few things but any gains made in one part of the image are undone by faults appearing in other parts.
Especially difficult are situations such as on the left building where we have lines numbers close to or even overlapping part of the design.
Here we see 2 examples of such situations
also note that font usage is not consistent between images.
It's worth noting that the lines are almost always obviously thinner then the fond used for the numbers so perhaps something could be done with that?
I have also tried using the EAST OCR system with the following code:
img = cv2.imread('vdb7C.jpg')
W=5664
H=4000
dim = (W, H)
img = cv2.resize(img, dim, interpolation = cv2.INTER_AREA)
net = cv2.dnn.readNet("frozen_east_text_detection.pb")
blob = cv2.dnn.blobFromImage(img, 1.0, (W, H),
(123.68, 116.78, 103.94), swapRB=True, crop=False)
net.setInput(blob)
(scores, geometry) = net.forward(["feature_fusion/Conv_7/Sigmoid",
"feature_fusion/concat_3"])
(numRows, numCols) = scores.shape[2:4]
rects = []
confidences = []
# loop over the number of rows
for y in range(0, numRows):
# extract the scores (probabilities), followed by the geometrical
# data used to derive potential bounding box coordinates that
# surround text
scoresData = scores[0, 0, y]
xData0 = geometry[0, 0, y]
xData1 = geometry[0, 1, y]
xData2 = geometry[0, 2, y]
xData3 = geometry[0, 3, y]
anglesData = geometry[0, 4, y]
for x in range(0, numCols):
if scoresData[x] < confidence:
continue
(offsetX, offsetY) = (x * 4.0, y * 4.0)
angle = anglesData[x]
cos = np.cos(angle)
sin = np.sin(angle)
h = xData0[x] + xData2[x]
w = xData1[x] + xData3[x]
endX = int(offsetX + (cos * xData1[x]) + (sin * xData2[x]))
endY = int(offsetY - (sin * xData1[x]) + (cos * xData2[x]))
startX = int(endX - w)
startY = int(endY - h)
rects.append((startX, startY, endX, endY))
confidences.append(scoresData[x])
boxes = non_max_suppression(np.array(rects), probs=confidences)
for box in boxes:
(y,h,x,w) = box
print(box)
print(np.shape(img))
cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.imwrite("output.jpg" , img)
however this causes quite a number of bounding boxes to be outside of the image and in general the bounding boxes seem unrelated to the content, so anyone know what's up there?
Any suggestions? I have 8000 images right now and need to eventually process a total of about 400k images.

I suggest using a solution that applies neural networks like keras-ocr which applies CRAFT and CRNN. It does a better job in detecting text that overlaps with the design. This is what I got using it out of the box:
import matplotlib.pyplot as plt
import keras_ocr
detector = keras_ocr.detection.Detector()
image = keras_ocr.tools.read('vdb7C.jpg')
boxes = detector.detect(images=[image])[0]
canvas = keras_ocr.tools.drawBoxes(image, boxes)
plt.imshow(canvas)
Result:

Run your tesseract piece of code, but only use results with 3 or more digits. This should provide you with enough good examples of digits. Extract each digit to separate file and save their positions. Now you can go two ways.
You can go the simple way if you will see that the fonts of the digits are quite similar. Then you can create a set of templates for the digits (say 15-30). Remember that you can get the size of the digits for specific image? Resize your digits template to the right size and run the most trivial template matching. This will definitely create some false detections (especially for "1"s), and you will have to find a way to reduce their amount to acceptable level.
More complex way is to build a custom CNN detector and train it on your data. So from the first stage you will get several hundred examples of digits (and their positions) that you want to detect. You can look at this project or this one as references. Also this article can provide you some guidance.
One more thing that can be useful. Your images have lots of long perpendicular lines. If you align them to the axis, you can remove the lines very easily by binarizing the original, shifting the result (right or down) by several pixels and ANDing them. This will left only the long lines. Find their length, and you will be able to remove lines above certain length in the original image.

Related

OpenCV perspective transform

I am currently working on creating a python software that tracks players on a soccer field. I got the player detection working with YoloV3 and was able to output quite a nice result with players centroids and boxes drawn. What i want to do now is translate the players position and project their centroids onto a png/jpg of a soccerfield. For this I inteded to use two arrays with refrence points one for the soccerfield-image and one for the source video. But my question now is how do I translate the coordinates of the centroids to the soccerfield image.
Similiar example:
Example
How the boxes and Markers are drawn:
def draw_labels_and_boxes(img, boxes, confidences, classids, idxs, colors, labels):
# If there are any detections
if len(idxs) > 0:
for i in idxs.flatten():
# Get the bounding box coordinates
x, y = boxes[i][0], boxes[i][1]
w, h = boxes[i][2], boxes[i][3]
# Draw the bounding box rectangle and label on the image
cv.rectangle(img, (x, y), (x + w, y + h), (255, 255, 255), 2)
cv.drawMarker (img, (int(x + w / 2), int(y + h / 2)), (x, y), 0, 20, 3)
return img
Boxes are generated like this:
def generate_boxes_confidences_classids(outs, height, width, tconf):
boxes = []
confidences = []
classids = []
for out in outs:
for detection in out:
# print (detection)
# a = input('GO!')
# Get the scores, classid, and the confidence of the prediction
scores = detection[5:]
classid = np.argmax(scores)
confidence = scores[classid]
# Consider only the predictions that are above a certain confidence level
if confidence > tconf:
# TODO Check detection
box = detection[0:4] * np.array([width, height, width, height])
centerX, centerY, bwidth, bheight = box.astype('int')
# Using the center x, y coordinates to derive the top
# and the left corner of the bounding box
x = int(centerX - (bwidth / 2))
y = int(centerY - (bheight / 2))
# Append to list
boxes.append([x, y, int(bwidth), int(bheight)])
confidences.append(float(confidence))
classids.append(classid)
return boxes, confidences, classids
Assuming a stationary camera,
Find the coordinates of the four corners of the field.
Find the corresponding four corners in the top-view image that you want to create.
Find a homography matrix using these two sets of points. You can use OpenCV's findHomography for this.
Transform all the centroids using this homography matrix and that should give you your coordinates in the new image space. You can use warpPerspective for doing this.
Recently during COVID19 pandemic many developers have developed "social-distancing-monitoring-system". There a few of them also developed "Bird's Eye View" of the system. Your problem is just similar. As external links are not accepted here, so I am not able to post the exact link(s). Please check their codes in GitHub.

Set an ROI in OpenCV

I am very new to OpenCV and Python. I have followed a tutorial to use YOLO using yolov3-tiny. It can detect vehicles fine. But what I need to complete my project is to count the number of vehicles that passes a particular lane. If I use the method where the vehicle is detected (the bounding box appears) to count, the count becomes very inaccurate since, the bounding box keeps blinking (meaning, it keeps on locating the same vehicle again, sometimes up to 5 times), so this is not a good way to count.
So I figured, how about if I just count a vehicle if it gets to a certain point. I have seen a lot of codes that seems to make this but, since I am a beginner, it really is hard for me to understand let alone, run it in my system. Their samples need to install so many things that I can't do because it throws errors.
See my sample code below:
cap = cv2.VideoCapture('rtsp://username:password#xxx.xxx.xxx.xxx:xxx/cam/realmonitor?channel=1')
whT = 320
confThreshold = 0.75
nmsThreshold = 0.3
list_of_vehicles = ["bicycle","car","motorbike","bus","truck"]
classesFile = 'coco.names'
classNames = []
with open(classesFile, 'r') as f:
classNames = f.read().rstrip('\n').split('\n')
modelConfiguration = 'yolov3-tiny.cfg'
modelWeights = 'yolov3-tiny.weights'
net = cv2.dnn.readNetFromDarknet(modelConfiguration, modelWeights)
net.setPreferableBackend(cv2.dnn.DNN_BACKEND_OPENCV)
net.setPreferableTarget(cv2.dnn.DNN_TARGET_CPU)
total_vehicle_count = 0
def getVehicleCount(boxes, class_name):
global total_vehicle_count
dict_vehicle_count = {}
if(class_name in list_of_vehicles):
total_vehicle_count += 1
# print(total_vehicle_count)
return total_vehicle_count, dict_vehicle_count
def findObjects(ouputs, img):
hT, wT, cT = img.shape
bbox = []
classIds = []
confs = []
for output in outputs:
for det in output:
scores = det[5:]
classId = np.argmax(scores)
confidence = scores[classId]
if confidence > confThreshold:
w, h = int(det[2] * wT), int(det[3] * hT)
x, y = int((det[0] * wT) - w/2), int((det[1] * hT) - h/2)
bbox.append([x, y, w, h])
classIds.append(classId)
confs.append(float(confidence))
indices = cv2.dnn.NMSBoxes(bbox, confs, confThreshold, nmsThreshold)
for i in indices:
i = i[0]
box = bbox[i]
getVehicleCount(bbox, classNames[classIds[i]])
x, y, w, h = box[0], box[1], box[2], box[3]
cv2.rectangle(img, (x,y), (x+w, y+h), (255,0,255), 1)
cv2.putText(img, f'{classNames[classIds[i]].upper()} {int(confs[i]*100)}%', (x,y-10), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (255,0,255), 2)
while True:
success, img = cap.read()
blob = cv2.dnn.blobFromImage(img, 1/255, (whT,whT), [0,0,0], 1, crop=False)
net.setInput(blob)
layerNames = net.getLayerNames()
outputnames = [layerNames[i[0]-1] for i in net.getUnconnectedOutLayers()]
# print(outputnames)
outputs = net.forward(outputnames)
findObjects(outputs, img)
cv2.imshow('Image', img)
if cv2.waitKey(1) & 0XFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
With this code, 1 vehicle is counted sometimes up to 50 count, depending on the size, which is highly inaccurate. How can I create an ROI so that when the detected vehicle passes that point, that will be the only time it will count.
First, I would recommend that you consider using a visual tracker to track each detected rectangle. This is important even if you have an ROI to crop the image close to your counting zone/line. That is because even if the ROI is localized, the detection might still blink a couple of times causing a miscount. That is especially valid if another vehicle can enter the ROI while the first one is still passing it.
I recommend using the easy-to-use tracker provided by the widely used dlib library. Please refer to this example on how to use it.
Instead of counting detections within an ROI, you need to define an ROI line (within your ROI). Then, track detections rectangles centers in each frame. Finally, increase your counter once a rectangle center passes the ROI line.
Regarding how to count a rectangle passing the ROI line:
Select two points to define your ROI line.
Use your points to find the parameters for the general line formula ax + by + c = 0
For each frame, plug the rectangle center coordinates in the formula and keep track of the sign of the result.
If the sign of the result changes that means that the rectangle center has passed the line.

How to delete or clear contours from image?

I'm working with license plates, what I do is apply a series of filters to it, such as:
Grayscale
Blur
Threshhold
Binary
The problem is when I doing this, there are some contour like this image at borders, how can I clear them? or make it just black color (masked)? I used this code but sometimes it falls.
# invert image and detect contours
inverted = cv2.bitwise_not(image_binary_and_dilated)
contours, hierarchy = cv2.findContours(inverted,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
# get the biggest contour
biggest_index = -1
biggest_area = -1
i = 0
for c in contours:
area = cv2.contourArea(c)
if area > biggest_area:
biggest_area = area
biggest_index = i
i = i+1
print("biggest area: " + str(biggest_area) + " index: " + str(biggest_index))
cv2.drawContours(image_binary_and_dilated, contours, biggest_index, [0,0,255])
center, size, angle = cv2.minAreaRect(contours[biggest_index])
rot_mat = cv2.getRotationMatrix2D(center, angle, 1.)
#cv2.warpPerspective()
print(size)
dst = cv2.warpAffine(inverted, rot_mat, (int(size[0]), int(size[1])))
mask = dst * 0
x1 = max([int(center[0] - size[0] / 2)+1, 0])
y1 = max([int(center[1] - size[1] / 2)+1, 0])
x2 = int(center[0] + size[0] / 2)-1
y2 = int(center[1] + size[1] / 2)-1
point1 = (x1, y1)
point2 = (x2, y2)
print(point1)
print(point2)
cv2.rectangle(dst, point1, point2, [0,0,0])
cv2.rectangle(mask, point1, point2, [255,255,255], cv2.FILLED)
masked = cv2.bitwise_and(dst, mask)
#cv2_imshow(imgg)
cv2_imshow(dst)
cv2_imshow(masked)
#cv2_imshow(mask)
Some results:
The original plates were:
Good result 1
Good result 2
Good result 3
Good result 4
Bad result 1
Bad result 2
Binary plates are:
Image 1
Image 2
Image 3
Image 4
Image 5 - Bad result 1
Image 6 - Bad result 2
How can I fix this code? only that I want to avoid that bad result or improve it.
INTRODUCTION
What you are asking starts to become complicated, and I believe there is not anymore a right or wrong answer, just different ways to do this. Almost all of them will yield positive and negative results, most likely in a different ratio. Having a 100% positive result is quite a challenging task, and I do believe my answer does not reach it. Yet it can be the basis for a more sophisticated work towards that goal.
MY PROPOSAL
So, I want to make a different proposal here.
I am not 100% sure why you are doing all the steps, and I believe some of them could be unnecessary.
Let's start from the problem: you want to remove the white parts on the borders (which are not numbers).
So, we need an idea about how to distinguish them from the letters, in order to correctly tackle them.
If we just try to contour and warp, it is likely to work on some images and not on others, because not all of them look the same. This is the hardest problem to have a general solution that works for many images.
What are the difference between the characteristics of the numbers and the characteristics of the borders (and other small points?):
after thinking about that, I would say: the shapes! That meaning, if you would imagine a bounding box around a letter/number, it would look like a rectangle, whose size is related to the image size. While in the case of the border, they are usually very large and narrow, or too small to be considered a letter/number (random points).
Therefore, my guess would be on segmentation, dividing the features via their shape. So we take the binary image, we remove some parts using the projection on their axes (as you correctly asked in the previous question and I believe we should use) and we get an image where each letter is separated from the white borders.
Then we can segment and check the shape of each segmented object, and if we think these are letters, we keep them, otherwise we discard them.
THE CODE
I wrote the code before as an example on your data. Some of the parameters are tuned on this set of images, so they may have to be relaxed for a larger dataset.
import cv2
import matplotlib.pyplot as plt
import numpy as np
%matplotlib inline
import scipy.ndimage as ndimage
# do this for all the images
num_images = 6
plt.figure(figsize=(16,16))
for k in range(num_images):
# read the image
binary_image = cv2.imread("binary_image/img{}.png".format(k), cv2.IMREAD_GRAYSCALE)
# just for visualization purposes, I create another image with the same shape, to show what I am doing
new_intermediate_image = np.zeros((binary_image.shape), np.uint8)
new_intermediate_image += binary_image
# here we will copy only the cleaned parts
new_cleaned_image = np.zeros((binary_image.shape), np.uint8)
### THIS CODE COMES FROM THE PREVIOUS ANSWER:
# https://stackoverflow.com/questions/62127537/how-to-clean-binary-image-using-horizontal-projection?noredirect=1&lq=1
(rows,cols)=binary_image.shape
h_projection = np.array([ x/rows for x in binary_image.sum(axis=0)])
threshold_h = (np.max(h_projection) - np.min(h_projection)) / 10
print("we will use threshold {} for horizontal".format(threshold))
# select the black areas
black_areas_horizontal = np.where(h_projection < threshold_h)
for j in black_areas_horizontal:
new_intermediate_image[:, j] = 0
v_projection = np.array([ x/cols for x in binary_image.sum(axis=1)])
threshold_v = (np.max(v_projection) - np.min(v_projection)) / 10
print("we will use threshold {} for vertical".format(threshold_v))
black_areas_vertical = np.where(v_projection < threshold_v)
for j in black_areas_vertical:
new_intermediate_image[j, :] = 0
### UNTIL HERE
# define the features we are looking for
# this parameters can also be tuned
min_width = binary_image.shape[1] / 14
max_width = binary_image.shape[1] / 2
min_height = binary_image.shape[0] / 5
max_height = binary_image.shape[0]
print("we look for feature with width in [{},{}] and height in [{},{}]".format(min_width, max_width, min_height, max_height))
# segment the iamge
labeled_array, num_features = ndimage.label(new_intermediate_image)
# loop over all features found
for i in range(num_features):
# get a bounding box around them
slice_x, slice_y = ndimage.find_objects(labeled_array==i)[0]
roi = labeled_array[slice_x, slice_y]
# check the shape, if the bounding box is what we expect, copy it to the new image
if roi.shape[0] > min_height and \
roi.shape[0] < max_height and \
roi.shape[1] > min_width and \
roi.shape[1] < max_width:
new_cleaned_image += (labeled_array == i)
# print all images on a grid
plt.subplot(num_images,3,1+(k*3))
plt.imshow(binary_image)
plt.subplot(num_images,3,2+(k*3))
plt.imshow(new_intermediate_image)
plt.subplot(num_images,3,3+(k*3))
plt.imshow(new_cleaned_image)
that produces the output (in the grid, left image are the input images, central one are the images after the mask based on histogram projections, and on the right are the cleaned images):
CONCLUSIONS:
As said above, this method does not yield 100% positive results. The last picture has lower quality and some parts are unconnected, and they are lost in the process. I personally believe this is a price to pay to get cleaner image, and if you have a lot of images, it won't be a problem, and you can remove those kind of images. Overall, I think this method returns quite clear images, where all other parts that are not letters or numbers are correctly removed.
ADVANTAGES
the image is clean, nothing more than letters or numbers are kept
the parameters can be tuned, and should be consistent across images
in case of problem, using some prints or some debugging on the loop that chooses the features to keep should make it easier to understand where are the problem and correct them
LIMITATIONS
it may fail in some cases where letters and numbers touch the white borders, which seems quite possible. It is handled from the black_areas created using the projection, but I am not so confident this will work 100% of the time.
some small parts of the numbers can be lost during the process, as in the last picture.

How to find table like structure in image

I have different type of invoice files, I want to find table in each invoice file. In this table position is not constant. So I go for image processing. First I tried to convert my invoice into image, then I found contour based on table borders, Finally I can catch table position.
For the task I used below code.
with Image(page) as page_image:
page_image.alpha_channel = False #eliminates transperancy
img_buffer=np.asarray(bytearray(page_image.make_blob()), dtype=np.uint8)
img = cv2.imdecode(img_buffer, cv2.IMREAD_UNCHANGED)
ret, thresh = cv2.threshold(img, 127, 255, 0)
im2, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
margin=[]
for contour in contours:
# get rectangle bounding contour
[x, y, w, h] = cv2.boundingRect(contour)
# Don't plot small false positives that aren't text
if (w >thresh1 and h> thresh2):
margin.append([x, y, x + w, y + h])
#data cleanup on margin to extract required position values.
In this code thresh1, thresh2 i'll update based on the file.
So using this code I can successfully read positions of tables in images, using this position i'll work on my invoice pdf file. For example
Sample 1:
Sample 2:
Sample 3:
Output:
Sample 1:
Sample 2:
Sample 3:
But, now I have a new format which doesn't have any borders but it's a table. How to solve this? Because my entire operation depends only on borders of the tables. But now I don't have a table borders. How can I achieve this? I don't have any idea to move out from this problem. My question is, Is there any way to find position based on table structure?.
For example My problem input looks like below:
I would like to find its position like below:
How can I solve this?
It is really appreciable to give me an idea to solve the problem.
Thanks in advance.
Vaibhav is right. You can experiment with the different morphological transforms to extract or group pixels into different shapes, lines, etc. For example, the approach can be the following:
Start from the Dilation to convert the text into the solid spots.
Then apply the findContours function as a next step to find text
bounding boxes.
After having the text bounding boxes it is possible to apply some
heuristics algorithm to cluster the text boxes into groups by their
coordinates. This way you can find a groups of text areas aligned
into rows and columns.
Then you can apply sorting by x and y coordinates and/or some
analysis to the groups to try to find if the grouped text boxes can
form a table.
I wrote a small sample illustrating the idea. I hope the code is self explanatory. I've put some comments there too.
import os
import cv2
import imutils
# This only works if there's only one table on a page
# Important parameters:
# - morph_size
# - min_text_height_limit
# - max_text_height_limit
# - cell_threshold
# - min_columns
def pre_process_image(img, save_in_file, morph_size=(8, 8)):
# get rid of the color
pre = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# Otsu threshold
pre = cv2.threshold(pre, 250, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
# dilate the text to make it solid spot
cpy = pre.copy()
struct = cv2.getStructuringElement(cv2.MORPH_RECT, morph_size)
cpy = cv2.dilate(~cpy, struct, anchor=(-1, -1), iterations=1)
pre = ~cpy
if save_in_file is not None:
cv2.imwrite(save_in_file, pre)
return pre
def find_text_boxes(pre, min_text_height_limit=6, max_text_height_limit=40):
# Looking for the text spots contours
# OpenCV 3
# img, contours, hierarchy = cv2.findContours(pre, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
# OpenCV 4
contours, hierarchy = cv2.findContours(pre, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
# Getting the texts bounding boxes based on the text size assumptions
boxes = []
for contour in contours:
box = cv2.boundingRect(contour)
h = box[3]
if min_text_height_limit < h < max_text_height_limit:
boxes.append(box)
return boxes
def find_table_in_boxes(boxes, cell_threshold=10, min_columns=2):
rows = {}
cols = {}
# Clustering the bounding boxes by their positions
for box in boxes:
(x, y, w, h) = box
col_key = x // cell_threshold
row_key = y // cell_threshold
cols[row_key] = [box] if col_key not in cols else cols[col_key] + [box]
rows[row_key] = [box] if row_key not in rows else rows[row_key] + [box]
# Filtering out the clusters having less than 2 cols
table_cells = list(filter(lambda r: len(r) >= min_columns, rows.values()))
# Sorting the row cells by x coord
table_cells = [list(sorted(tb)) for tb in table_cells]
# Sorting rows by the y coord
table_cells = list(sorted(table_cells, key=lambda r: r[0][1]))
return table_cells
def build_lines(table_cells):
if table_cells is None or len(table_cells) <= 0:
return [], []
max_last_col_width_row = max(table_cells, key=lambda b: b[-1][2])
max_x = max_last_col_width_row[-1][0] + max_last_col_width_row[-1][2]
max_last_row_height_box = max(table_cells[-1], key=lambda b: b[3])
max_y = max_last_row_height_box[1] + max_last_row_height_box[3]
hor_lines = []
ver_lines = []
for box in table_cells:
x = box[0][0]
y = box[0][1]
hor_lines.append((x, y, max_x, y))
for box in table_cells[0]:
x = box[0]
y = box[1]
ver_lines.append((x, y, x, max_y))
(x, y, w, h) = table_cells[0][-1]
ver_lines.append((max_x, y, max_x, max_y))
(x, y, w, h) = table_cells[0][0]
hor_lines.append((x, max_y, max_x, max_y))
return hor_lines, ver_lines
if __name__ == "__main__":
in_file = os.path.join("data", "page.jpg")
pre_file = os.path.join("data", "pre.png")
out_file = os.path.join("data", "out.png")
img = cv2.imread(os.path.join(in_file))
pre_processed = pre_process_image(img, pre_file)
text_boxes = find_text_boxes(pre_processed)
cells = find_table_in_boxes(text_boxes)
hor_lines, ver_lines = build_lines(cells)
# Visualize the result
vis = img.copy()
# for box in text_boxes:
# (x, y, w, h) = box
# cv2.rectangle(vis, (x, y), (x + w - 2, y + h - 2), (0, 255, 0), 1)
for line in hor_lines:
[x1, y1, x2, y2] = line
cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)
for line in ver_lines:
[x1, y1, x2, y2] = line
cv2.line(vis, (x1, y1), (x2, y2), (0, 0, 255), 1)
cv2.imwrite(out_file, vis)
I've got the following output:
Of course to make the algorithm more robust and applicable to a variety of different input images it has to be adjusted correspondingly.
Update: Updated the code with respect to the OpenCV API changes for findContours. If you have older version of OpenCV installed - use the corresponding call. Related post.
You can try applying some morphological transforms (such as Dilation, Erosion or Gaussian Blur) as a pre-processing step before your findContours function
For example
blur = cv2.GaussianBlur(g, (3, 3), 0)
ret, thresh1 = cv2.threshold(blur, 150, 255, cv2.THRESH_BINARY)
bitwise = cv2.bitwise_not(thresh1)
erosion = cv2.erode(bitwise, np.ones((1, 1) ,np.uint8), iterations=5)
dilation = cv2.dilate(erosion, np.ones((3, 3) ,np.uint8), iterations=5)
The last argument, iterations shows the degree of dilation/erosion that will take place (in your case, on the text). Having a small value will results in small independent contours even within an alphabet and large values will club many nearby elements. You need to find the ideal value so that only that block of your image gets.
Please note that I've taken 150 as the threshold parameter because I've been working on extracting text from images with varying backgrounds and this worked out better. You can choose to continue with the value you've taken since it's a black & white image.
There are many types of tables in the document images with too much variations and layouts. No matter how many rules you write, there will always appear a table for which your rules will fail. These types of problems are genrally solved using ML(Machine Learning) based solutions. You can find many pre-implemented codes on github for solving the problem of detecting tables in the images using ML or DL (Deep Learning).
Here is my code along with the deep learning models, the model can detect various types of tables as well as the structure cells from the tables: https://github.com/DevashishPrasad/CascadeTabNet
The approach achieves state of the art on various public datasets right now (10th May 2020) as far as the accuracy is concerned
More details : https://arxiv.org/abs/2004.12629
this would be helpful for you.
I've drawn a bounding box for each word in my invoice, then I will chose only fields that I want. You can use for that ROI (Region Of Interest)
import pytesseract
import cv2
img = cv2.imread(r'path\Invoice2.png')
d = pytesseract.image_to_data(img, output_type=pytesseract.Output.DICT)
n_boxes = len(d['level'])
for i in range(n_boxes):
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
img = cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 1)
cv2.imshow('img', img)
cv2.waitKey(0)
You will get this output:

Text recognition and detection using TensorFlow

I a working on a text recognition project.
I have built a classifier using TensorFlow to predict digits but I would like to implement a more complex algorithm of text recognition by using text localization and text segmentation (separating each character) but I didn't find an implementation for those parts of the algorithms.
So, do you know some algorithms/implementation/tips I, using TensorFlow, to localize text and do text segmentation in natural scenes pictures (actually localize and segmentation of text in the scoreboard for sports pictures)?
Thank you very much for any help.
To group elements on a page, like paragraphs of text and images, you can use some clustering algo, and/or blob detection with some tresholds.
You can use Radon transform to recognize lines and detect skew of a scanned page.
I think that for character separation you will have to mess with fonts. Some polynomial matching/fitting or something. (this is a very wild guess for now, don't take it seriously).
But similar aproach would allow you to get the character out of the line and recognize it in same step.
As for recognition, once you have a character, there is a nice trigonometric trick of comparing angles of the character to the angles stored in a database.
Works great on handwriting too.
I am not an expert on how page segmentation exactly works, but it seems that I am on my way to become one. Just working on a project including it.
So give me a month and I'll be able to tell you more. :D
Anyway, you should go and read Tesseract code to see how HP and Google did it there. It should give you pretty good ideas.
Good luck!
After you are done with Object Detection, you can perform text detection which can be passed on to tesseract. There can multiple variation to enhance image before passing it to detector function.
Reference Papers
https://arxiv.org/abs/1704.03155v2
https://arxiv.org/pdf/2002.07662.pdf
def text_detector(image):
#hasFrame, image = cap.read()
orig = image
(H, W) = image.shape[:2]
(newW, newH) = (640, 320)
rW = W / float(newW)
rH = H / float(newH)
image = cv2.resize(image, (newW, newH))
(H, W) = image.shape[:2]
layerNames = [
"feature_fusion/Conv_7/Sigmoid",
"feature_fusion/concat_3"]
blob = cv2.dnn.blobFromImage(image, 1.0, (W, H),
(123.68, 116.78, 103.94), swapRB=True, crop=False)
net.setInput(blob)
(scores, geometry) = net.forward(layerNames)
(numRows, numCols) = scores.shape[2:4]
rects = []
confidences = []
for y in range(0, numRows):
scoresData = scores[0, 0, y]
xData0 = geometry[0, 0, y]
xData1 = geometry[0, 1, y]
xData2 = geometry[0, 2, y]
xData3 = geometry[0, 3, y]
anglesData = geometry[0, 4, y]
# loop over the number of columns
for x in range(0, numCols):
# if our score does not have sufficient probability, ignore it
if scoresData[x] < 0.5:
continue
# compute the offset factor as our resulting feature maps will
# be 4x smaller than the input image
(offsetX, offsetY) = (x * 4.0, y * 4.0)
# extract the rotation angle for the prediction and then
# compute the sin and cosine
angle = anglesData[x]
cos = np.cos(angle)
sin = np.sin(angle)
# use the geometry volume to derive the width and height of
# the bounding box
h = xData0[x] + xData2[x]
w = xData1[x] + xData3[x]
# compute both the starting and ending (x, y)-coordinates for
# the text prediction bounding box
endX = int(offsetX + (cos * xData1[x]) + (sin * xData2[x]))
endY = int(offsetY - (sin * xData1[x]) + (cos * xData2[x]))
startX = int(endX - w)
startY = int(endY - h)
# add the bounding box coordinates and probability score to
# our respective lists
rects.append((startX, startY, endX, endY))
confidences.append(scoresData[x])
boxes = non_max_suppression(np.array(rects), probs=confidences)
for (startX, startY, endX, endY) in boxes:
startX = int(startX * rW)
startY = int(startY * rH)
endX = int(endX * rW)
endY = int(endY * rH)
# draw the bounding box on the image
cv2.rectangle(orig, (startX, startY), (endX, endY), (0, 255, 0), 3)
return orig

Categories

Resources