Disparity Estimation Method Based on an Improved ACV Model

Fig. 4. Sceneflight disparity maps separately incorporating multi-scale feature fusion network. (a) Bonsai prediction; (b) motorcycle prediction; (c) bicycle prediction

Download full size

Fig. 5. Sceneflow disparity maps of the whole network. (a) Headphone cable prediction; (b) bonsai and fine line prediction; (c) barrel texture prediction

Download full size

Fig. 6. KITTI disparity maps of the whole network. (a) Pedestrian prediction; (b) roadside lamppost prediction; (c) pedestrian prediction

Download full size

Table 1. Only the ablation analysis of the attention branch trained on Sceneflow dataset
View table
Table 1. Only the ablation analysis of the attention branch trained on Sceneflow dataset
Method EPE /pixel D1 /%
ACV 0.798 2.91
ACV+Sobel 0.691 2.54
ACV+multi_fea_fusion 0.786 2.87
ACV+Sobel+multi_fea_fusion （MS-ACV） 0.687 2.54

Table 2. Only the ablation analysis of parallax branches trained on Sceneflow dataset
View table
Table 2. Only the ablation analysis of parallax branches trained on Sceneflow dataset
Method EPE /pixel D1 /%
ACV 0.718 2.55
ACV+Sobel 0.627 2.23
ACV+multi_fea_fusion 0.664 2.29
ACV+Sobel+multi_fea_fusion（MS-ACV） 0.605 2.14

Table 3. The ablation analysis of the entire network trained on Sceneflow dataset
View table
Table 3. The ablation analysis of the entire network trained on Sceneflow dataset
Method EPE /pixel D1 /%
ACV 0.484 1.60
ACV+Sobel 0.472 1.53
ACV+multi_fea_fusion 0.479 1.57
ACV+Sobel+multi_fea_fusion（MS-ACV） 0.467 1.51

Table 4. Comparative analysis of the entire network trained on KITTI dataset

View table

Table 4. Comparative analysis of the entire network trained on KITTI dataset

Method	KITTI12						KITTI15
Method	2-Noc /%	2-All /%	3-Noc /%	3-All /%	EPE Noc /pixel	EPE All /pixel	D1-bg /%	D1-fg /%	D1-All /%
GC-Net	2.71	3.46	1.77	2.30	0.6	0.7	2.21	6.16	2.87
PSMNet	2.44	3.01	1.49	1.89	0.5	0.6	1.86	4.62	2.32
EdgeStereo	2.32	2.88	1.46	1.83	0.4	0.5	1.84	3.30	2.08
GWCNet	2.16	2.71	1.32	1.70	0.5	0.5	1.74	3.93	2.11
ACVNet	1.83	2.35	1.13	1.47	0.4	0.5	1.37	3.07	1.65
MS-ACV	1.75	2.30	1.11	1.44	0.4	0.5	1.32	3.08	1.61

Tools

Get Citation

Copy Citation Text

Lunming Qin, Bin Yu, Haoyang Cui, Houqin Bian, Xi Wang. Disparity Estimation Method Based on an Improved ACV Model[J]. Laser & Optoelectronics Progress, 2024, 61(24): 2415003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Machine Vision

Received: Mar. 6, 2024

Accepted: Apr. 25, 2024

Published Online: Dec. 19, 2024

The Author Email:

DOI:10.3788/LOP240835

CSTR:32186.14.LOP240835

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Only the ablation analysis of the attention branch trained on Sceneflow dataset

Table 1. Only the ablation analysis of the attention branch trained on Sceneflow dataset

Table 2. Only the ablation analysis of parallax branches trained on Sceneflow dataset

Table 2. Only the ablation analysis of parallax branches trained on Sceneflow dataset

Table 3. The ablation analysis of the entire network trained on Sceneflow dataset

Table 3. The ablation analysis of the entire network trained on Sceneflow dataset

Table 4. Comparative analysis of the entire network trained on KITTI dataset

Table 4. Comparative analysis of the entire network trained on KITTI dataset