comparing text in sequential images

[jjanderson5]

Hi,

I am a total newbie to image processing and ImageMagick. I am a programmer with experience in C, C++, java and python, but have programmed in non-image based applications. I am working on a new web based application and have images containing text passing through the application. The nice thing is that the images are always the same size and in the same location and the text tends to be in the same location too. The bad thing, for me the programmer, is that I do not actually have the text available as text, and so I need an image processor.

So what am I looking for?

1) be able to extract text from localized areas
2) be able to compare 2 sequential images to determine what has changed in localized areas
3) must be able too both in less than 0.25 seconds

Question 1 - Is this doable with ImageMagick?

I assume the answer is, 'yes', but better to find out early if I am trying to do something unrealistic.

Question 2 - Where do I look for more information? (are there tutoritials? readme files? API documentation?)

If someone can point me in the right direction, I will be most appreciative.

Jim A.

[jjanderson5]

After posting, I walked the dog and while walking I realized there is one other thing that I am looking for.

3) I need to be able to capture images from the screen display.

I could use pointers on how to do this also.

Jim A.

[snibgo]

ImageMagick does not contain OCR (Optical Character Recognition) functionality.

[jjanderson5]

Thank you. That is good to know.

As part of the API, is there an ability to read the pixels from an image?

Jim

[snibgo]

Certainly, and much more. There are many APIs. I use the C API, known as MagickCore, to directly read and write pixels.

It seems your main requirement is to convert images to text. IM can't do that. It might be useful when preparing images for OCR.

[Bonzo]

You could use compare to see the difference between images but I do not know if that will give you the information you want. Speed would probably depend on the type and size of the input images.