CODEWITHSUNDEEP
c#.netimage-processingopencvemgucv
kingUD
kingUD

Reputation: 21

Object Recognition using SURF in emgu CV

For my project I need to recognize the Object(In my case it is a door) using SURF. I'm using emguCV (openCv c# wrapper).

I have been working with the surf feature detection example from the emguCV library.

By using this I can recognize only few number of object image (e.g.: door) with the given template (model) image.

Since there is different kind of doors, for some door images it was not detected as the door. So, I decided If I can put different kind of door images (2 or more images) as model(template)image, then I can get more accurate result.

My problem is I have no clear idea about how to write a code for work with 2 or more model (template) images? Basically, I want to know by using which parameter, we can say that it is detected or not from the code?

(Visually if it is detected, it draws a red rectangle around the matched object.)

Will be really grateful for your help.

Thank you

Upvotes: 2

Views: 8408

Answers (1)

Taimoor
Taimoor

Reputation: 11

i did this in my last semester as a course project in Data Structures. In that i am trying to detect a face from a cam. I have worked a lot on that in the start then after 3,4 months i was able to complete that finally.. It detects that faces properly. It brings the green line around the face when detected. The methodology is the same as for the face or some doors or the cars or anything else. The question here is that how to get started? Yes, it is easy, you just need to use EMGU CV and OPEN CV (as i did). I must tell here that EMGU CV is for .Net and same way OPEN CV is for some other. The only thing you have to look for is the "XML of the door" that's it!. rest of the libraries are easily available here: http://sourceforge.net/projects/emgucv/ you can download it from the preceding link. What next? Next is you have to just add these libraries. When done, you can use then as according to your requirement. As i mentioned above you if u are detecting doors then you need the "XML of the door". Because the XML which you use in your code is defining that detection. because XML tells that which object you need to detect, in my example i used XML of face to detect face. You can use for the door. You have one more option to whether pick a XML file of a door from the net or whether you can make changes to the existing one you have. And as you asked to use 2 or more doors images, this approach is not the professional one!!. Lets say if the image of a door that you did not hard coded before and a new door comes then what? Nothing will happen. no detection then! Rest is on your requirement. I hope its a little clear now that what to look for now!!.. Regards Taimoor Alam.

Upvotes: 1

Related Questions