Need suggestion on how to take clear closeup shot with cheap camera to OCR

[jimdagys]

I have a cheap digital camera that does not have any focus adjustment. I want to use the camera (advantage: much more portable than a scanner) in lieu of a scanner to capture printed pages and then use an OCR program to convert to text. The problem is, because there is no focus adjustment, the print is not very clear (see screenshot) and so the OCR program has some difficulty. I have already tried different lighting settings, etc. I am wondering if I could hold some kind of cheap lens (such as magnifying glass) in front of the camera so that I can get a clear closeup shot. Any suggestions?

[Ethlred]

Back up about a foot. Fixed focus cameras usually claim to be in focus from 3 feet to infinity. Thats false. They usually are in tolerable focus at around 4 or 5 feet on out. Depending on the lens aperture.

Borrow a camera that focuses.

Avoid on the camera flash. You may already figured this one out. A diffuser can help if you must use a flash.

A hand-held magnifying glass is not going to cut it.

If by some stroke of insanity (an insanely great fixed focus camera from the mind of Steven Jobs for instance) the manufacturer put in threads for filters buy a set of close up filters. They come or at least used to come as a set of three. Usually +1, +2 and +4 diopters. They can be stacked together in various ways to get up to +7 diopters.

If there is no threading you could try tape. Try to get a close fit as the filters MUST fit flat. If you get them out of alignment you will change the angle of the focal plane and you are not going to have much depth of field if that camera has fixed aperture as well as a fixed focus. Then again most fixed focus FILM cameras had small apertures to increase the depth of field. Unfortunately, the close up lenses will decrease your depth of field. This a matter of physics and not quality. Close up photography has inherent depth of focus issues.

To add to problem cheap cameras don't have a flat field of focus in the first place and you NEED a flat field for a sheet of paper. Good macro lenses for close up work with flat objects are usually long focus despite the added cost. The longer distance flattens the field. That is, it is farther to the edge of the paper than to the center from the camera's field of focus. The farther away you have the camera the less this matter of geometry gets in the way.

Course none of that finicky stuff matters in this case. You just need enough quality for the OCR to do its work. And you can't do any of that without a MUCH more expensive setup.

For lighting its best to have two lights. One on each side of the camera. At least try to get a diffuse source from the side. A large window on the north side of building for instance. Shoot on a white background as the wood is showing through. If the paper is printed on both sides then a black background would be a better choice. Black might be better in any case as it cuts down on reflections. If any of the stuff you want to shoot is glossy you may even need to mask the camera in black but it is not likely you would need to do that for OCR.

[jimdagys]

Thanks for the tips. The exact above blurry picture of text yields the following when fed directly to Abbyy Finereader ver8. I'm curious if anybody has another OCR program (such as Adobe), if you could process the above image and compare the results to the following Abbyy OCR.

English Newspaper Reading Examination A.
Answer the following questions;
1 Newspapers can he defined in many wayt. Name 5 newspaper definitions.
% Name 5 personal] associated with the operation of a newspaper.
% Name 5 dinfereni sections which can be found in a newspaper.
4 What is the quickest way to gunge the theme of a newspaper story?
5. Name 3 w ays of reading a newspaper article quickly.
6 Name 2 Chinese newspapers which you would be able to read in English.
Name 3 foreign newspapers which you would be able to read in English.
§ Name 5 foreign magazine* which you would be able to read in English.
9. there arc many words which arc quite often repeatedly used in various newspaper
stone*. In your own words define the following 5 terms:
, aV Agenda
b Bottom Line
C. Context
d Dimensions
c Establishment
10. What would you expect to find in the obituary section of a newspaper?
11. Newspapers ad ways have a crossword puzzle. What is a crossword puzzle?
12* Name 4 ways in which you can contact a newspaper.
13. Name 2 reasons why people read newspapers?
14. Which section of the newspaper accepts public notices?
15. Read the following newspaper story entitled, "Walking on My Eardrums."
In your own words & using only one sentence, summarize this story
16 Choose the most appropriate headline for the following 2 newspaper stories.
Story I.
a Criminals in Syria
b. Lebanon*! Prime Minister: Syria is Threatening My Country
c. UN in Lebanon
Story 2.
a. Silverwater Jail
b. A Morning Walk
c. Mcneill Faces Patton Murder Charge

[ahhaa]

JD-
you have several probs going here-

> your photo is not straight- this makes it more difficult for the recog software

> your contrast is very low- many free photo programs have a one click enhancement button

> you are not filling the frame with text- you can do this by getting closer or cropping the frame.

All the pixels are considered by the software whether relevant or not, excess just becomes overhead.

Many photopaint programs have edge enhancement too.

You can get a film camera closeup lens for peanuts these days, and you could take a tip from the old Minox spy camera- they used a short chain that would establish the minimum focus distance.

Brighter light closes the f-stop, increasing depth of field.


All that said, you are copying exam papers? You've just left a permanent record on the internet...

[zoobie]

10. What would you expect to find in the obituary section of a newspaper?

Fried chicken?