Table 4.

Quantitative measurement of food items based on point cloud completion and the ICP algorithm.

Food Object Item	Ground Truth ${}^{\mathit{a}}$ (cm${}^3$)	Estimated Volume ${}^{\mathit{b}}$ ${\mathit{V}}_{\mathit{g}}$ (cm${}^3$)	SD of ${\mathit{V}}_{\mathit{g}}$ (cm${}^3$)	Error of ${\mathit{V}}_{\mathit{g}}$${}^{\mathit{c}}$ (%)	Estimated Volume ${}^{\mathit{d}}$ ${\mathit{V}}_{\mathit{p}}$ (cm${}^3$)	SD of ${\mathit{V}}_{\mathit{p}}$ (cm${}^3$)	Error of ${\mathit{V}}_{\mathit{p}}$ e (%)
1—Banana	-	181.6	2.8	-	175.6	12.6	3.3
2—Orange	203.0	199.5	1.2	1.7	186.7	12.1	8.0
3—Pear	-	192.75	0.9	-	175.5	12.8	8.9
4—Cube	185.2	187.0	0.8	1.0	198.6	19.3	7.2
5—Potted Meat Can	-	369.0	4.5	-	381.0	24.5	3.3
6—Lemon	-	98.5	1.3	-	91.0	3.4	7.6
7—Tuna Fish Can	180.0	176.0	4.2	2.2	193.3	8.3	7.3
8—Pudding Box	342.0	358.5	13.2	4.8	309.7	3.5	9.4

${}^a$ The reference volumes of the food items. ${}^b$ The mean estimated volume based on point cloud completion using known extrinsic parameters. ${}^c$ The error of the estimated volume using point cloud completion. ${}^d$ The mean estimated volume using the iterative closest point algorithm without prior knowledge. ^e The error of the estimated volume using the iterative closest point algorithm.