Pick out object on image (or video)

[ST-Peterburg]

Hello Friends!

prompt me please!

I want to "teach" my VI to pick out object on image (or video). My VI must to read information (text) from some documents with OCR function. But I must to create ROI myself and my VI works very bad because documents may move and ROI become wrong. So I want to my VI'll find text itself and set ROI on this text. Is it able?

Any help and example will thanked!

[ST-Peterburg]

I know function - Pattern Matching and Geometric Matching. But I don't know how to use this function for text, because text is different on documents. 

[muks]

You can use "imaq select rectangle" vi to select the roi. Can you post a few images?

[ST-Peterburg]

It is a Drivers licence. And I want to my VI find the text and set ROI on it. Imaq select rectangle - is it the function and can I find it in Vision Assistant?

[muks]

You can use a clamp also and determine the position of the card. After this you apply the relative roi to read the text using ocr

[ST-Peterburg]

Thanks! I have made clamp already, but it is no comfortable for me. I have solved to do my VI more perfection.

How I can to set ROI myself when snap with camera has done already? What is the function or functions do it possible?

[muks]

As suggested earlier did you try "IMAQ select rectangle".vi?

[ST-Peterburg]

I am trying now. Thanks for idea!

[Andrey_Dmitriev · Accepted solution]

Something like that:

First detect two edges of the driving license, then compute intersection between two edges, then set coordinate system, and after that reposition ROI in OCR step depend from coordinate system:

Now OCR work also for rotated image:

Take a note, that Straight Edge (Rake) will works for 8 bit images. Because RGB is not supported you will need to convert your color image to grayscale.

Script in attachment.

Andrey.

[ST-Peterburg]

Thank You, Andrey! I'll try to do as You recommended

.