Draw a rectangle around a string of words using pytesseract-CodePudding

Here is my image:

I can recognize the words:

I need to check if there is a line with specific text in the image and highlight this line with a rectangle. For example. I check if there is "times, it was the worst". And then I expect to see:

How can I achieve this?

My code:

import cv2 as cv
from pytesseract import pytesseract, Output
unchanged_image = cv.imread("i1Abv.png", cv.IMREAD_UNCHANGED)
initial_image = cv.imread("i1Abv.png", 0)
cv.imshow('', initial_image)
cv.waitKey(0)

ret, image = cv.threshold(initial_image, 100, 255, cv.THRESH_BINARY)

cv.imshow('', image)
cv.waitKey(0)

results = pytesseract.image_to_data(image, output_type=Output.DICT, config="--psm 6")
for i in range(0, len(results["text"])):
    # extract the bounding box coordinates of the text region from
    # the current result
    x = results["left"][i]
    y = results["top"][i]
    w = results["width"][i]
    h = results["height"][i]
    # extract the OCR text itself along with the confidence of the
    # text localization
    text = results["text"][i]
    conf = float(results["conf"][i])

    # filter out weak confidence text localizations
    if conf > 10:

        # strip out non-ASCII text, so we can draw the text on the image
        # using OpenCV, then draw a bounding box around the text along
        # with the text itself
        cv.rectangle(unchanged_image, (x, y), (x   w, y   h), (0, 255, 0), 2)
        cv.putText(img=unchanged_image, text=text, org=(x, y), fontFace=cv.FONT_HERSHEY_COMPLEX,
                   fontScale=0.3, color=(36, 0, 255), thickness=1)

cv.imshow('', unchanged_image)
cv.waitKey(0)

CodePudding user response：

pytesseract.image_to_data provides line_num for each recognized text block. You can group all the recognitions by line_num and then concatenate words into a text line. Also you need to find a bounding box that includes all the text boxes of each word (you can get them using left, top, width, height). You can find it with

Code:

import cv2
import numpy as np
import pytesseract

original_image = cv2.imread("text_line.png")
gray = cv2.cvtColor(original_image, cv2.COLOR_BGR2GRAY)
target_text = "times, it was the worst"

df = pytesseract.image_to_data(gray, lang="eng", config="--psm 6", output_type=pytesseract.Output.DATAFRAME)

# group recognized words by lines
for line_num, words_per_line in df.groupby("line_num"):
    # filter out words with a low confidence
    words_per_line = words_per_line[words_per_line["conf"] >= 5]
    if not len(words_per_line):
        continue

    words = words_per_line["text"].values
    line = " ".join(words)
    print(f"{line_num} '{line}'")

    if target_text in line:
        print("Found a line with specified text")
        word_boxes = []
        for left, top, width, height in words_per_line[["left", "top", "width", "height"]].values:
            word_boxes.append((left, top))
            word_boxes.append((left   width, top   height))

        x, y, w, h = cv2.boundingRect(np.array(word_boxes))
        cv2.rectangle(original_image, (x, y), (x   w, y   h), color=(255, 0, 255), thickness=3)

cv2.imwrite("out.jpg", original_image)