Robotic Visual Tracking of Relevant Cues in Underwater Environments with Poor Visibility Conditions

Table 1

Comparison of the five most relevant regions detected as relevant or interesting (nonwater regions) by using the NVT and the proposed method. In the last row the total number of images and the average of the percentage of detected relevant regions are specified.

Seq.	Frames	Depth [m]	% of interesting regions (NVT)	% of interesting regions (AVA)
1	168	7.7	75.20	95.85
2	243	7.8	66.91	95.80
3	153	7.8	83.00	99.60
4	181	11.8	57.79	92.15
5	163	7.1	74.47	98.28
6	242	11.3	59.92	90.76
	1150		69.04	94.90