CT Image Denoising with Non-Local Means Based on Feature Fusion

Chao Long; Heng Jin; Ling Li; Jinyin Sheng; Liming Duan

doi:10.3788/AOS202242.1134024

Acta Optica Sinica, Volume. 42, Issue 11, 1134024(2022)

CT Image Denoising with Non-Local Means Based on Feature Fusion

Chao Long^1,2, Heng Jin^1,3, Ling Li^1,3, Jinyin Sheng^1,3, and Liming Duan^1,2、*

¹Engineering Research Center of Industrial Computed Tomography Nondestructive Testing, Ministry of Education, Chongqing University, Chongqing 400044, China

²College of Optoelectronic Engineering, Chongqing University, Chongqing 400044, China

³College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. Schematic diagram of NLM algorithm

Download full size

Fig. 2. Flowchart of algorithm

Download full size

Fig. 3. Gray level curves of Sheep-Logan head model under different similarity windows. (a) Sheep-Logan head model; (b) size of similarity window is 2×2, and size of search window is 5×5; (c) size of similarity window is 3×3, and size of search window is 7×7; (d) size of similarity window is 4×4, and size of search window is 9×9

Download full size

Fig. 4. Denoising results and local magnification of Sheep-Logan head model under different noise. (a) Original image; (b) δ=5; (c) δ=10; (d) δ=15

Download full size

Fig. 5. Gray contrast curves

Download full size

Fig. 6. CT slices of sheep bone. (a) Original image; (b) local enlargement; (c) NLM; (d) ST-NLM; (e) proposed algorithm

Download full size

Fig. 7. CT images of 168^th slice of sheep bone. (a) Original image; (b) local enlargement; (c) NLM; (d) ST-NLM; (e) proposed algorithm

Download full size

Fig. 8. CT images of 105^th slice of rock sample and results of edge detection. (a) Original image; (b) local enlargement; (c) NLM; (d) ST-NLM; (e) proposed algorithm

Download full size

Fig. 9. CT images of 179^th slice of rock sample and edge detection results. (a) Original image; (b) local enlargement;(c) NLM; (d) ST-NLM; (e) proposed algorithm

Download full size

Table 1. Comparison of evaluation indexes of simulation image

View table

Table 1. Comparison of evaluation indexes of simulation image

δ	Noise in Fig. 4(a)	S_SIMof NLM	S_SIM of ST-NLM	S_SIM of proposed method
5	0.4203	0.5193	0.5333	0.5593
10	0.3114	0.4843	0.4961	0.525
15	0.2527	0.4759	0.4818	0.5016
δ	Noise in Fig. 4(a)	P_SNRof NLM	P_SNR of ST-NLM	P_SNR of proposed method
5	24.0114	28.5805	30.1507	32.1923
10	19.3119	24.9168	26.7773	29.5706
15	16.8237	24.9743	25.9465	27.6429

Table 2. Main parameters of CT system
View table
Table 2. Main parameters of CT system
Object CT system X-ray energy /keV X-ray energy /μA Detector size /mm Exposure time /ms
Bone of sheep Planer array 60 40 0.2 200
Stone Planer array 150 65 0.2 1000

Table 3. Indexes of CT image sharpness

View table

Table 3. Indexes of CT image sharpness

Image	Evaluation parameter	Slice No.	Sharpness
Image	Evaluation parameter	Slice No.	NLM	ST-NLM	This paper
		12^th	76.7816	77.1844	82.3463
Fig. 6	Tenengrad function	25^th	52.4985	53.3265	59.3399
		196^th	65.2964	66.4159	80.8070
Fig. 7	Tenengrad function	168^th	25.5605	25.6018	28.4846

Table 4. Sharpness of CT image of rock sample
View table
Table 4. Sharpness of CT image of rock sample
Image Evaluation parameter Slice No. Sharpness
NLM ST-NLM This paper
Fig. 8 Tenengrad function 105^th 23.9975 24.0052 25.1846
Fig. 9 Tenengrad function 179^th 26.6517 27.1457 28.7604

Tools

Get Citation

Copy Citation Text

Chao Long, Heng Jin, Ling Li, Jinyin Sheng, Liming Duan. CT Image Denoising with Non-Local Means Based on Feature Fusion[J]. Acta Optica Sinica, 2022, 42(11): 1134024

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites