Sorry, but it seems you think the power factor is just cos phi, where phi is the angle that current is shifted in relation to the voltage.

That is only true for perfect sine waves.

In reality the power factor is calculated by measuring both current and voltage and sample-by-sample compute the product of the current and voltage readings.

1) you may have overtones on the curves so they are not sinusoidal.

2) if the load is a cheap unit with half-wave rectifier then you may have a load current for one half-period and no load on the other half-period.

That means that you should add a known DC voltage to your input signals (U and I) and let the ADC measure this reference voltage so it knows how many ticks of offset you have. Or you have to make sure that you have two identical resistors subdividing the voltage reference in which case you may be able to just assume that the offset is 50% of the ADC range.

In your case, it isn't possible to just use a software highpass filter, since the current curve can be very non-symmetric so filtering can't find out the zero value.

See this thread:
http://www.8052.com/forumchat/read/164714