Skip to main content
. Author manuscript; available in PMC: 2019 Oct 12.
Published in final edited form as: ACM Trans Knowl Discov Data. 2013 Sep;7(3):10.

Fig. 19.

Fig. 19.

(top-left) All scaling subsequences created from the query Q scaled to different lengths from 80 to 100. (top-right) The upper envelope U and lower envelope L created by the maximum and minimum values of all scaling subsequences in (left). The lower bound distance is computed from the candidate C to the closer envelope, as shown in vertical lines. (bottom) The query is scaled from 60 to 100, or has a scaling factor of 40%. Note that although the scaling factor may be large, in most cases, the lower bound is still large and can prune candidates efficiently.