There are at least 36 ways to separate blurred areas from non-blurred ( take a look at this survey : http://optica.csic.es/papers/icpr2k.pdf ).


Of course, another way you could be judging the blurriness of an image is the ratio of the power of the high frequencies relative to the lower frequencies. In this case, instead of obtaining the standard deviation of each MxN window, you would have to obtain the 2D FFT. This is slightly more complicated though.