Why tesseract is failing to give better output even though the text is very clear?

Question

I have to extract text from some screenshot images like this Screenshot.png

1. First I have to detect and crop the powershell from the screenshot like this (used yolov5 to detect) Cropped.JPEG . after croping the image i am saving it as JPEG.

2. Then i have to extract text from this cropped image.

I know for tesseract the background should be light and the text should be dark so i inverted the image inverted image.JPEG . After doing this, for few images i am getting expected output but for most of the images i am not getting desired output.

i have tried following methods.

1. Resize
2. Thresholding
3. Grayscal
4. Adding borders

I am not sure what i am doing wrong...

Answer 1

If you are using pytessaract then there is no need for thresholding. just resizing and conversion to grayscale will be enough to get the desired output.

gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
text = pytesseract.image_to_string(gray, lang='eng')
text_to_arr = (text.split('\n'))
space_to_empty = [x.strip() for x in text_to_arr]
space_clean_list = [x.lower() for x in space_to_empty if x]