The best OCR tools available for personal use in terms of accuracy are Nuance OmniPage, ABBYY FineReader and Acrobat XI Pro. (I have not looked at benchmarks but I got this impression when I was looking for one a few years ago). But they all cost over 20 USD.
Tesseract is most of the time considered as being the most accurate open source OCR engine. There exist several Tesseract GUIs. Otherwise you can use some bindings to it, as in the example below.
Example:
Convert image to tif with ImageMagick:
#convert myimage.jpeg -auto-level -compress none myimage.tif
Python code to read data from myimage.tif (Tesseract needs TIFF as input):
from PIL import Image
from pytesser.pytesser import *
image_file = 'myimage.tif'
im = Image.open(image_file)
text = image_to_string(im)
text = image_file_to_string(image_file)
text = image_file_to_string(image_file, graceful_errors=True)
print "=====output=======\n"
print text
FYI: comparison of optical character recognition software