Acta Optica Sinica, Volume. 40, Issue 5, 0510002(2020)
Depth Estimation Method of Light Field Image Based on Occlusion Scene
Fig. 1. Imaging model of Lytro camera
Fig. 2. Diagram of 4D light field
Fig. 3. Multi-view image
Fig. 4. Digital refocusing principle
Fig. 5. Effect comparison diagram. (a) Constrained adaptive defocus depth map; (b) traditional defocus depth map
Fig. 6. Defocus depth map and consistent depth map. (a) Constrained adaptive defocus depth map; (b) constrained angle entropy depth map
Fig. 7. Map results. (a) Initial depth map; (b) confidence map; (c) edge detection result; (d) final depth estimation result
Fig. 8. Depth estimation results of MonasRoom light field reference dataset. (a) MonasRoom central perspective image; (b) GT value; (c) the proposed method; (d) method in Ref. [13]; (e) method in Ref.[8]; (f) method in Ref.[5]
Fig. 9. Depth estimation results of Budden light field reference dataset. (a) Budden central perspective image; (b) GT value; (c) the proposed method; (d) method in Ref.[13]; (e) method in Ref.[8]; (f) method in Ref.[5]
Fig. 10. Depth estimation results of 5×5 view light field images taken by Lytro
|
Get Citation
Copy Citation Text
Xiaomin Liu, Mengzhu Du, Zhibang Ma, Yunfei Zhu, Pengbo Chen, Fengying Ma. Depth Estimation Method of Light Field Image Based on Occlusion Scene[J]. Acta Optica Sinica, 2020, 40(5): 0510002
Category: Image Processing
Received: Sep. 3, 2019
Accepted: Nov. 18, 2019
Published Online: Mar. 10, 2020
The Author Email: Liu Xiaomin (liuxmamara@zzu.edu.cn)