Vision-Based Satellite Recognition and Pose Estimation Using Gaussian Process Regression

Table 1

Satellite recognition results on BUAA-SID 1.0 and comparison with [17, 19]. Results are in bold if the proposed method performs the best.

Representation	Method	Accuracy (%)
HU	Kernel method [19]; Our GPR method	54.13
		71.00
FD	Kernel method [19]; Our GPR method	57.52
		62.30
HOG	Kernel method [19]; Our GPR method	98.00
		86.61
MI+FD+region covariance [17]+HOG	KLPP [17]	95.87
	LPP [17]	88.22
	KPCA [17]	69.70
	PCA [17]	68.57