Help to translate captcha by using the python, csv and Tesseract language [closed]

Question

I’m trying to make the recognition of these images, I’ve tried several types and I can’t identify with the tesserat because of the lines and scratches

I get the site information below and can’t translate using cv2 and Tesseract I used the code below for testing, but no success

if I need more examples, I pick up on this site https://afi.santanderfinanciamentos.com.br/scripts/afi_santanderLojista.dll?OPERA=AFI_Home

Captchas used

--Code that I used--

img = cv2.imread('Imagens/solvCaptch.png')
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
suave = cv2.GaussianBlur(img, (3, 3), 0)
(T, bin) = cv2.threshold(suave, 160, 255, cv2.THRESH_BINARY)
resultado = np.vstack([np.hstack([bin])])
str = ocr.image_to_string(resultado, lang='por')
print(str)