Acta Optica Sinica, Volume. 41, Issue 11, 1133002(2021)
Color Constancy with Multi-Channel Confidence-Weighted Method
Fig. 1. Network structure
Fig. 2. Flow chart of the color constancy algorithm
Fig. 3. Ambiguous of the light source estimation
Fig. 4. Comparison of the number of confidence-weighted areas of single-channel and three-channel light source
Fig. 5. CC module network structure diagram
Fig. 6. Samples of the dataset
Fig. 7. Diagram of the network training stage
Fig. 8. Graph of the network training loss value
Fig. 9. Visualization of the network output. (a) Input images; (b) using proposed algorithm to estimate images corrected by light source; (c) weight distribution diagrams of proposed algorithm to weight the image multi-channel confidence; (d) standard light source; (e) Grey-world algorithm; (f) White-Patch algorithm; (g) Shades-of-Grey algorithm; (h) Grey-Edge algorithm
|
|
|
Get Citation
Copy Citation Text
Zepeng Yang, Kai Xie, Tong Li, Mengyao Yang, Bin Yang. Color Constancy with Multi-Channel Confidence-Weighted Method[J]. Acta Optica Sinica, 2021, 41(11): 1133002
Category: Vision, Color, and Visual Optics
Received: Dec. 4, 2020
Accepted: Jan. 18, 2021
Published Online: Jun. 7, 2021
The Author Email: Xie Kai (2596898130@qq.com)