@bradfordcross That cited paper didn't impress me at all. Look at Fig. 3 which should show segments with correlation and w/o.