Hi

As a matter of fact, it is. Of course, you might apply more optimization techniques to that ready-to-use embedded function to work better. Because you may need some more elaborated labelled features, whereas bwlabel() might not be handle such level of complexity. Hence, you can either write a new function for more acquisition of efficiency or apply some processing schemes to build that labelled stuffs as you want.

For example, that method is probably unable to label the neuro-fuzzy features for noise-free channels.

Good luck