how to calculate performance of standard edge based algorithm in MATLAB?

hello, I want to compare performance of standard edge based algorithm i.e.(Robert, Canny ,Sobel ,LOG,Prewitt) with each other.Can any one sugggest code for it. I have done my implementation on standard edge based algorithm but facing problem in parameter chosen for comparing it? Is it possible to implement normal probablity density function or normal density function in standard edge based algo?if it is ,so please suggest me what to do?

Answer 1

[Image Analyst]

We don't know what you mean by "normal probablity density function or normal density function". What would the distribution be of ?

I don't know what the "standard edge based algorithm" is. They all use different algorithms. The "true" location of an edge is often a judgment call, so who's to say which on is "right"? What counts is if you can get your task done after using one of them.

So whatyou do is to just try them all, tweaking each until you achieve the "best" edges then see if you can complete the rest of the task. If you have ground truth then you can see which one did best. You can use an ROC curve perhaps. Maybe one didn't give closed curves and you needed closed curves to get solid blobs. So that one would not find as many as some other edge algorithm. You can compare them that way.

[Prerna Surbhi]

Thanks for advise, I have done edge based image segmentation based on its algorithm.the algorithm which are involved in this implementation are 'sobel','Robert','canny','Prewitt' ,'LOg' ,'zero cross'.i am attaching my implementation below.In this implementation i have found that,'Canny' is the best edge based segmentation algorithm.which covers almost all edge. so i want to express it by how much more more percent or value canny method covers the edge .in compare to all other method.

<<

>>