MATLAB Answers

problem regarding edge detection

7 views (last 30 days)
matlab_image on 15 Aug 2012
I am applying edge detection methods like Sobel, prewitt,robert etc. Can anybody please tell me how can I find probability of detection & Probability of False alarm with different values of SNR, after applying these edge detection methods... Thanks in advance..


Sign in to comment.

Answers (1)

Image Analyst
Image Analyst on 15 Aug 2012
Call bwlabel to label the images. Then use something like ginput to ask your user which edges are genuine and which are false. Then just take the ratio of the number of false blobs divided by the total number of blobs.


Show 4 older comments
matlab_image on 20 Aug 2012
Thanks a lot......I have 3 questions to ask from you: 1)Can any of you please explain it a little bit more.I am still unable to calculate the value of PD and PFA... Please help me I need it very urgent.... I have applied bwlabel and ginput but after this how to calculate the ratio of false blobs and total blobs.................. 2)when I click on false edges how can i know how many times user clicked for false edge..Is it taken manually or is there any method for counting no. of clicks...............3)I have another problem which is not related with this is that I am facing lots of problems while logging into and opening Mathworks. Can you please tell me what is the reason behind this...
Image Analyst
Image Analyst on 20 Aug 2012
1) If you don't know if an edge is a true edge or a false edge then I can't help you. Perhaps you can just make up random numbers in that case. I mean, if you don't know what's true and false then who would?
2) numberOfClicks = length(x); where x is returned from ginput().
3) Call the Mathworks and tell them that you can't use their web site because of some kind of a login problem.
matlab_image on 22 Aug 2012
Thanks for answering..... I know which are false edges and which are true.. but I have doubts about PD and PFA..... So I will try to clear these while reading some tutorials........ Thanks again....

Sign in to comment.

Sign in to answer this question.