A Novel Multiscale Edge Detection Approach Based on Nonsubsampled Contourlet Transform and Edge Tracking

Table 2

The and values, where the 1, 2, 3, and 4 denote top, middle top, middle bottom, and bottom original images in Figure 12(a), respectively.

Original images		Comparison to other methods
		Canny	LoG	Sobel	M-Sobel	Our method
1		0.4626	0.5626	0.8452	0.8588	0.9444
		0.4064	0.4691	0.8193	0.8377	0.8351
2		0.2672	0.3240	0.5829	0.6384	0.6271
		0.2046	0.2726	0.5376	0.6243	0.5685
3		0.4474	0.5038	0.6418	0.6560	0.6832
		0.3716	0.5968	0.7227	0.7690	0.7310
4		0.7468	0.7149	0.8484	0.8500	0.8740
		0.6459	0.6839	0.8527	0.8807	0.7792