激光与光电子学进展, 2020, 57 (18): 181027, 网络出版: 2020-09-02
边缘复杂光场图像的深度估计散焦响应函数优化 下载: 1003次
Defocusing Response Function Optimization in Depth Estimation of Boundary Complex Light Field Image
图 & 表
图 1. 散焦结合相关性深度估计算法流程
Fig. 1. Flowchart of depth estimation algorithm based on defocus and correspondence
图 2. 二阶导算子比较图。(a)传统拉普拉斯算子;(b)本文提出的能量增强的二阶导算子
Fig. 2. Comparison of second-order derivative operators. (a) Traditional Laplace operator; (b) proposed second-order derivative operator for energy enhancement
图 3. 基于"shoe"图像的深度估计结果对比。(a)光场原图像;(b) DCDC算法获取的深度图;(c)本文方法获取的深度图
Fig. 3. Comparison of depth evaluation results obtained based on "shoe" image. (a) Light field raw image; (b) depth map calculated by DCDC algorithm; (c) depth map calculated by proposed method
图 4. 散焦响应曲线对比。(a)基于拉普拉斯算子的散焦响应曲线图;(b)基于本文能量增强二阶导算子的散焦响应曲线
Fig. 4. Comparison of defocus response curves. (a) Defocus response curves obtained based on Laplace operator; (b) defocus response curves obtained based on proposed energy enhanced second-order derivative operator
图 5. 针对"stapler"图像的深度评价结果对比。(a)去冗余光场原图像;(b) DCDC算法获取的深度图;(c)本文方法获取的深度图
Fig. 5. Comparison of depth evaluation results for "stapler" image. (a) Light field raw image without redundancy; (b) depth map calculated by DCDC algorithm; (c) depth map calculated by proposed method
图 6. benchmark数据集中的2个样本图。(a) "boxes"场景;(b) "dino"场景
Fig. 6. Two sample images in benchmark dataset. (a) "boxes" scene; (b) "dino" scene
图 7. 针对"boxes"场景的深度评价结果对比。(a)真实深度图;(b) DCDC算法获取的深度图;(c)本文方法获取的深度图
Fig. 7. Comparison of depth evaluation results for "boxes" scene. (a) Ground truth; (b) depth map calculated by DCDC algorithm; (c) depth map calculated by proposed method
表 1深度图精度的定量评价
Table1. Quantitative evaluation of accuracy of depth maps
|
武迎春, 程星, 谢颖贤, 王安红. 边缘复杂光场图像的深度估计散焦响应函数优化[J]. 激光与光电子学进展, 2020, 57(18): 181027. Yingchun Wu, Xing Cheng, Yingxian Xie, Anhong Wang. Defocusing Response Function Optimization in Depth Estimation of Boundary Complex Light Field Image[J]. Laser & Optoelectronics Progress, 2020, 57(18): 181027.