This seems like it's vulnerable to noisy signals creating spurious zero-crossings. If you have some high-amplitude harmonics, isn't it possible for them to create several crossings in a few samples as the wave for the fundamental tone is near zero?