光场图像透视变换算法 下载: 1191次
Perspective Transformation Algorithm for Light Field Image
1 江南大学数字媒体学院, 江苏 无锡 214122
2 江苏省媒体设计与软件技术重点实验室, 江苏 无锡 214122
图 & 表
图 1. 光场图像透视变换模型
Fig. 1. Perspective transformation model for light field image
下载图片 查看原文
图 2. 带空洞的光场图像透视变换结果。(a)输入的中心子视点图像;(b)单子视点图像透视变换结果;(c)全部子视点图像透视变换融合结果
Fig. 2. Perspective transformation results of light field image with black holes. (a) Input central subaperture view image; (b) perspective transformation result of single subaperture view image; (c) perspective transformation fusion result of all subaperture view images
下载图片 查看原文
图 3. 基于视差优先级的光场图像修复结果。(a)待修复的子视点图像;(b)待修复的子视点视差图;(c)修复后的子视点图像;(d)修复后的子视点视差图
Fig. 3. Light field image restoration results based on disparity priority. (a) Subaperture view image to be restored; (b) subaperture disparity map before restoration; (c) restored subaperture view image; (d) restored subaperture disparity map
下载图片 查看原文
图 4. 光场图像所有子视点的修复顺序
Fig. 4. Restoration order of all subaperture views in light field image
下载图片 查看原文
图 5. 光场图像透视变换及修复结果。(a1)~(f1)输入中心子视点图像;(a2)~(f2)透视变换结果;(a3)~(f3)修复后的透视变换结果
Fig. 5. Results of light field perspective transformation and restoration. (a1)-(f1) Input central subaperture view images; (a2)-(f2) perspective transformation results; (a3)-(f3) restored perspective transformation results
下载图片 查看原文
图 6. 光场图像透视变换结果比较。(a1)~(c1)文献[
8]算法结果;(a2)~(c2)文献[
9]算法结果;(a3)~(c3)本文算法的结果;(a4)~(c4)图像真值
Fig. 6. Comparison of light field image perspective transformation results. (a1)-(c1) Results of method in Ref. [8]; (a2)-(c2) results of method in Ref. [9]; (a3)-(c3) results of proposed method; (a4)-(c4) truth images
下载图片 查看原文
表 1透视变换参数
Table1. Parameters for perspective transformation
Parameter | Fig. 5(a) | Fig. 5(b) | Fig. 5(c) | Fig. 5(d) | Fig. 5(e) | Fig. 5(f) |
---|
b /mm | 1.00 | 1.00 | 1.50 | 1.12 | 1.65 | 0.86 | Tx /mm | -20 | -20 | -30 | -60 | -45 | -50 | Ty /mm | 0 | 0 | 0 | 0 | 0 | 0 | Tz /mm | 0 | 0 | 0 | 0 | 0 | 0 | Rx /(°) | 0 | 0 | 0 | 1 | 0.5 | 0.5 | Ry /(°) | 0 | 0.5 | 0.5 | 1 | 0.5 | 0.5 | Rz /(°) | 0 | 0 | 0 | 5 | -3 | -3 |
|
查看原文
表 2算法消耗时间
Table2. Consuming time of proposed algorithm
Parameter | Fig. 5(a) | Fig. 5(b) | Fig. 5(c) | Fig. 5(d) | Fig. 5(e) | Fig. 5(f) |
---|
T / s | 283.98 | 286.95 | 305.79 | 302.98 | 298.11 | 282.24 |
|
查看原文
表 3光场透视变换结果定量分析
Table3. Quantitative analysis for light field perspective transformation images
Parameter | Method | Fig. 6(a) | Fig. 6(b) | Fig. 6(c) |
---|
| Method in Ref. [8] | 23.18 | 22.38 | 22.72 | PSNR | Method in Ref. [9] | 26.35 | 22.06 | 22.53 | | Proposed | 35.43 | 25.69 | 30.74 | | Method inRef. [8] | 0.67 | 0.81 | 0.79 | SSIM | Method in Ref. [9] | 0.78 | 0.81 | 0.79 | | Proposed | 0.96 | 0.89 | 0.95 |
|
查看原文
王建明, 毛一鸣, 晏涛, 刘渊. 光场图像透视变换算法[J]. 激光与光电子学进展, 2019, 56(15): 151003. Jianming Wang, Yiming Mao, Tao Yan, Yuan Liu. Perspective Transformation Algorithm for Light Field Image[J]. Laser & Optoelectronics Progress, 2019, 56(15): 151003.