Selection of Visual Descriptors for the Purpose of Multi-camera Object Re-identification - Feature Selection for Data and Pattern Recognition - page 286

Fig. 12.17 Internal stability SI (the lower the better) calculated for each dataset separately

Fig. 12.18 Value-based ranking RV (the higher the better) for DSIM measure only with local image
descriptors included

0.6

Internal stability SI External stability SE

0.5

0.4

0.3

0.2

0.1

0.0

CLD CLD

Trans

EHD CMSP Hist HistFull LBPHist Moment

CentNorm

Moment
InvGPSO

VertTrace Opponent
Sift

Opponent
Surf64

Sift Surf128 Surf64

Fig. 12.19 Internal stability SI and external stability SE (the smaller the better) for DSIM measure
only with local image descriptors included

Basing on the DSIM measure only, it is possible to compare local-based features
with other descriptors. Local-based features provide better results for VEHICLE
(they adopt leading positions). This may be caused by the fact that vehicles provide
more distinguishable locations for local feature extraction that are more or less con-