extract char location from pdf gives the wrong y corrdinates

Question

I use the pdfminer to get the location of chars from a pdf, but it seems to give the wrong y coordinates

when I convert the pdf to image the all the pixels are in the wrong place I understand the pdfminer flip the y coordinate so I extract the page height and do y' = height - y but it still miss in about 5 pixel

why is that?

code:

from pdfminer.layout import LAParams, LTTextBox, LTText, LTChar, LTAnno
from pdfminer.pdfpage import PDFPage
from pdfminer.pdfinterp import PDFPageInterpreter, PDFResourceManager
from pdfminer.converter import PDFPageAggregator

# Imports Searchable PDFs and prints x,y coordinates
fp = open('/tst.pdf', 'rb')
manager = PDFResourceManager()
laparams = LAParams()
dev = PDFPageAggregator(manager, laparams=laparams)
interpreter = PDFPageInterpreter(manager, dev)
pages = PDFPage.get_pages(fp)

for page in pages:
    print('--- Processing ---')
    interpreter.process_page(page)
    layout = dev.get_result()
    x, y, text = -1, -1, ''
    _,_, width, height = page.mediabox
    
    for textbox in layout:
        if isinstance(textbox, LTText):
            for line in textbox:
                for char in line:
                    # If the char is a line-break or an empty space, the word is complete
                    if isinstance(char, LTAnno) or char.get_text() == ' ':
                        if x != -1:
                            print('At %r is text: %s' % ((x, height- y), text))
                        x, y, text = -1, -1, ''
                    elif isinstance(char, LTChar):
                        text += char.get_text()
                        if x == -1:
                            x, y, = char.bbox[0], char.bbox[3]
    # If the last symbol in the PDF was neither an empty space nor a LTAnno, print the word here
    if x != -1:
        print('At %r is text: %s' % ((x, height -y), text))