help with the application on image processing and setup

Hello Everyone

I am trying to build an application in which i am detecting the circular objects with certain size, I am not able to use imfindcircles on those images, because imfind cirlces creates cirles at places even where i dont even need a circle(since the objects are quite similar, but different size). Please see the attached image file for the type of image. I need to find the circles that are only of specific size range.

Answer 1

[arun Dhillon]

The advise as posted by the image analyst, using imfindcircles

(by defining radius range) worked, but in addition i have to add two more additions to the image, The first one being histeq filter, and the scond one using the edge threshold. TO compute and detect the exact circles that i want.

Answer 2

[Image Analyst]

Have you tried imfindcircles()?

Or seen Steve's blog on watershed separation?

[Image Analyst]

What do you really need to know? Number of blobs with smooth boundaries? Since mushrooms will be smooth but the other debris will not be. How about area fraction? Area fraction seems like it would be a good metric to determine when to harvest the field.

If you really need circles, you might try to implement the paper below

Link to paper

A NEW EFFICIENT ELLIPSE DETECTION METHOD

Yonghong Xie

Department of Computer Science

University of Nevada,

Qiang Ji

Department of Electrical, Computer, and Systems Engineering

Reno Rensselaer Polytechnic Institute

1051-4651/02 $17.00 (c) 2002 IEEE

ABSTRACT

In this paper, we introduce a new method for ellipse

detection. This method takes the advantages of major axis

of an ellipse to find ellipse parameter fast and efficiently.

It only needs an one-dimensional accumulator array to

accumulate the length information for minor axis of the

ellipse. This method does not require the evaluation of the

tangents or curvatures of the edge contours, which are

generally very sensitive to noise working conditions. No

complicated mathematical computation is involved in the

implementation and the required computational storage

space is much cheaper compared to the current methods.

Experiments with both synthetic and real images indicate

the effectiveness of the proposed method.

Answer 3

[arun Dhillon]

In the 1051-4651/02 link for the paper, will i find the steps and complete algorithm that i maybe able to fit on for my application,, because the 'link to the paper' that you provided me above (the steps for ellipse detection by converting pixels to one dimensions and then further) were very complicated and i did not understood the math for the same. please edify me help ke for the same. I am just stuck with this step for my application

[Image Analyst]

I have not (yet) coded that up yet. I'm only part way through it and then I had to drop it and go on to other things. Sorry.

FAQ

[arun Dhillon]

okay

please let me know when you are done with the code; bcause i believe that, your method could be the solution to my problem. Ans please feel free to contact me, and if you think you can help me with my application, please contact me as well.

regards

Arun deep singh Dhillon

[e-mail]

[Image Analyst]

You could be waiting a very long time. I last touched it a few months ago, and before that like two years ago. No end in sight.