Visual Security Assessment via Saliency-Weighted Structure and Orientation Similarity for Selective Encrypted Images

Table 1

Performance comparison employing different saliency models: graph-based visual saliency (GBVS), context-aware visual saliency (CAVS), saliency estimation using region covariance (CovSal), saliency through adaptive whitening of color and scale features (AWS), and Boolean map saliency (BMS).

Metric	PLCC	SRCC	KRCC	RMSE	Time(s)
Without saliency	0.943	0.935	0.769	0.435	0.356
GBVS [36]	0.956	0.954	0.815	0.388	0.476
CAVS [37]	0.958	0.956	0.815	0.386	21.43
CovSal [38]	0.949	0.943	0.809	0.395	1.256
AWS [39]	0.953	0.947	0.817	0.396	0.789
BMS [40]	0.951	0.948	0.813	0.401	6.579