We present a novel pair-wise comparison based psychovisual and analytical framework for automatic measurement of motion activity in video sequences. In [1] we constructed a test-set of video segments and a ground truth, based on subjective tests with naïve subjects. We presented automatically extractable descriptors of motion activity computed from MPEG block motion vectors, based on different hypotheses about subjective perception of motion activity. We tested the average error performance of the descriptors. In this paper we test the performance of the descriptors against the ground truth using pair-wise comparison of video segments. We show that all the descriptors perform well and that the MPEG-7 motion activity descriptor, based on variance of motion vector magnitudes, is one of the best in performance. We determine the common limitations of the proposed lowlevel motion activity descriptors. We find that some of the proposed descriptors are significantly less susceptible to the aforementioned limitations.