Image Recognition

[Sena_Scott]

Hi, currently i installed with

1)National Instruments Vision Assistant 8.0 

2)National Instruments Vision Builder AI 2.5

3) LabView 8.0

I'm totally new to this vision module.. I'm using a USB PHILIPS camera SPC900NC/00.. I need to use the image i get from the camera to somehow control the movements of my robot.. Like how can i make the robot differentiate RED colour from other colours( Blue, Green, Yellow, etc) throught the image i obtained?? My project is somehow is to make a robot move autnomously in a serach and rescue mission..  I kinda blur of which programs i should used for this process out the 3 Programs i installed as named above.. 

The best will be i can make use of the "Functions" palette in LabView 8.0 to carry out the process of differentiating the colours.. Cause i already using Labview 8.0 for the movement of my robot, I need to integrate both the image recognition and the movement of the robots.. 

Please Get back to me ASAP.. Thanks Guys.. ^_^

[Ray.R]

A good way to start is actually with the Vision Assistant (builder).  Did you check the documentation that came with it or any of the examples?

It's been almost 2 years since I used the Assistant.  I typically write the code without it.  But the assistant will generate the code for you and it can be a start.

I will check to see if there are examples or discussion posted within this forum.  You may also check the Knowledge Base to do a search.

RayR

[Ray.R]

THis link describes the difference between the two packages:

http://forums.ni.com/ni/board/message?board.id=170&message.id=307552

Extracting color:  http://forums.ni.com/ni/board/message?board.id=170&message.id=207157

Creating LV vi (I remeber this from the vision course - worth taking):

http://forums.ni.com/ni/board/message?board.id=170&message.id=310492

Question: Do you know about IMAQs?

[Sena_Scott]

Hi,

     Thanks.. I'm started u sing Vision Assistant.. Roughly i know how it works already..  But i encounter another problem, when i wanted to use "Processing Functions >> Identification", A message POP-UP and indicates my image which is in "RGB format" is unsuitable for the function only image in "binary/8-bit format" is suitable..  Just a reminder.. the Camera i'm using is PHILIPS SPC900NC/00.. I was wondering is there any way i can change the format form RGB into binary/8-bit?? or do i need to change to another camera?

Btw, assume the above problem is solve.. For example, there is 2 doors .. One label "1" and the other label "2".. How do i make sure the camera will only recognise "2" ?? Do i have to acquire an image of a "2" before hand??

Thanks..

[Ray.R]

Hi Sena.

You should have a image convertion function available.  Convert RGB to 8-bit format.

In order to recognize, you can scan door 2 and use it to identify (compare) the correct door.  The easier way would be to acquire an image before hand.

There are other methods, but they are more lengthy to implement.  Comparing to a baseline is the easier way.  You can also select the region of interest for you comaprison, which should speed up the process and minimize errors.  Try to establish a point of reference.

RayR

[effety741]

Picitup is a recognized computer vision innovator. Its a very help full site for Computer Vision Solutions, If u want you can try to visit this link image recognition. I think it will be help for you to Image recogition & computer vision solutions.