Screenshots and OCR, over a network

[chris414]

I have a couple decisions to make and was wondering whether you could give me some advice:

I will be using one relatively fast computer (Core Two Duo 2.4Ghz, 4Gb RAM, Vista 64bit) as a server, and a bunch of old computers (think about 10 years old) as clients. Basically I have to be able to read small snippets of text off the client computer's screen, and have the data available on the server computer.

I'm not sure whether it is best to:
a) Send an image from the client to server over a LAN containing the snippet of text, and then use the server to do a bit of simple OCR on the text to get the string
b) Make the client do the OCR and only send the string to the server

I have to read text from the screen of each computer at the very least once a second (preferably twice a second), and so i am worried that either there will be lag when transferring over the network or if i try to get old PC's to do the OCR... what's your gut feeling?

Btw this has to be done in Java / Python

 

[guest2013-1]

First off, you're explaining what you suggest you do to read text off of the clients PC's. You're not explaining why

You'll find most of the time (people who know wtf they're doing) would ask you why you want to read text off of client's PC's because there are several thousand other solutions to the problem instead of the one you want to follow.

Please explain why. What the purpose of this will be and then we can suggest something

 

[guest2013-1]

btw, there will definitely be lag

first thought in my head is "keylogger"

 

[dequadin]

This sounds either a bit sinister or like a varsity project

 

[guest2013-1]

This sounds either a bit sinister or like a varsity project

Not really, companies have policies in place where they can monitor your PC at work. However, the way he wants to do it sounds a bit dodge. Like "I want to prove you're on facebook all day and minimize the window whenever I (the boss) walk by!" or some ****.

Keylogger if you want to know what is typed

Screenshot will take up network resources and consume loads of power from both PC's especially if you want to check every 500ms.

 

[dequadin]

Hooking into the Windows API, and saving a screenshot as JPG should take up that many resources, you're looking at about 150KB.

Of course the image degradation you get from the JPG, may impact the OCR. Now if you're only interested in certain portions of the screen those could be extracted first and then sent. So many question so few answers.

chris414 what are you trying to achieve?