Object Edge Detection Algorithm Based on Improved Canny Algorithm

Fig. 2. Shapes of different functions at the same scale. (a) Optimal function; (b) Cauchy edge detection function; (c) the first derivative of the Gaussian function

Download full size

Fig. 3. Spatial structures of the edge detection function.(a) (b) γ=0.5, W=1; (c) γ=1.1, W=1; (d) γ=0.5, W=2

Download full size

Fig. 4. Edge detection filters. (a) Unnormalized operator; (b) normalized operator

Download full size

Fig. 5. rice image and its x-direction edge results. (a) Original image; (b) unnormalized filter edge image; (c) normalized filter edge image

Download full size

Fig. 6. Performance indicators of noise reduction algorithms

Download full size

Fig. 7. Comparison of the effects of different noise reduction algorithms. (a) Noisy image; (b) adaptive median-filtered image; (c) median-filtered image

Download full size

Fig. 8. P-R curves of different algorithms on the BSDS500 dataset

Download full size

Fig. 9. Comparison of edge results of different algorithms. (a) Canny algorithm; (b) Canny-Cauchy algorithm; (c) algorithm from reference [16]; (d) Sobel algorithm

Download full size

Table 1. Optimal edge detection function at different scales

View table

Table 1. Optimal edge detection function at different scales

n	$x_{m a x}$	$a_{1}$	$a_{2}$	$a_{3}$	$a_{4}$	$α$	$ω$
1	1.4	-0.00361	0.00400	2.13381	-1.00400	2.95584	1.72579
2	1.6	0.00253	0.00686	1.51633	-1.00686	2.68633	2.19106
3	1.8	0.02537	-0.05680	3.59605	-0.94320	2.19275	0.47157
4	2.0	-0.04449	-0.04159	2.12745	-0.95841	1.83140	0.96353

Table 2. Theoretical performance parameters of the edge detection function

View table

Table 2. Theoretical performance parameters of the edge detection function

W	$x_{m a x}$	$Σ Λ$			$F_{P R}$
W	$x_{m a x}$	Opt	Gauss	Cauchy	Gauss	Cauchy
W = 1	0.15	4.21	0.921	1.464	1.000	0.996
	0.3	2.87	0.921	1.446	1.000	0.983
	0.5	2.13	0.921	1.404	1.000	0.955
	0.8	1.57	0.921	1.317	0.999	0.892
	1.0	1.33	0.920	1.251	0.998	0.840
	1.2	1.12	0.918	1.186	0.996	0.785
W = 2	1.4	1.72	0.921	1.348	1.000	0.915
	1.6	1.57	0.921	1.317	1.000	0.892
	1.8	1.45	0.921	1.284	0.995	0.867
	2.0	1.34	0.921	1.251	0.991	0.840

Table 3. Edge results of different algorithms on the BSDS500 dataset
View table
Table 3. Edge results of different algorithms on the BSDS500 dataset
Algorithm ODS OIS AP R50 FPS
Human 0.800 0.800
Canny 0.610 0.656 0.576 0.757 40
Canny-Cauchy 0.656 0.693 0.635 0.827 32
Algorithm of reference［16］ 0.599 0.643 0.566 0.737 38
Sobel 0.552 0.585 0.521 0.604 60

Tools

Get Citation

Copy Citation Text

Xinshan Yu, Xiangyin Meng, Tengfei Jin, Jinze Luo. Object Edge Detection Algorithm Based on Improved Canny Algorithm[J]. Laser & Optoelectronics Progress, 2023, 60(22): 2212002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites