用于聚焦型光场相机标定的棋盘角点检测算法 下载: 1153次
Checkerboard Corner Detection Algorithm for Calibration of Focused Plenoptic Camera
1 海军航空大学, 山东 烟台 264001
2 国防科技大学前沿交叉学科学院, 湖南 长沙 410073
3 国防科技大学计算机学院, 湖南 长沙 410073
图 & 表
图 1. 相机成像模型。 (a)主透镜成像示意图; (b)微透镜成像示意图
Fig. 1. Camera imaging model. (a) Imaging diagram of main lens; (b) imaging diagram of micro-lenses
下载图片 查看原文
图 2. 角点检测结果。(a)初始角点检测结果; (b)去除微透镜边缘角点后的结果; (c) RANSAC算法拟合后的候选角点;(d)优化后的投影角点
Fig. 2. Corner detection results. (a) Initial corner detection result; (b) result after removal of corners on edges of micro-lenses; (c) candidate corners fitted by RANSAC algorithm; (d) projected corners after optimization
下载图片 查看原文
图 3. 角点检测结果示例。(a)含有噪声的情况下; (b)图像含有较大镜头畸变的情况;(c)同一微透镜图像中存在多个角点的情况; (d)漏检的情况
Fig. 3. Examples of corner detection results. (a) Case with image noise; (b) case with image suffering from large lens distortion; (c) case with multiple corners in one micro-image; (d) case with missed detection
下载图片 查看原文
图 4. 仿真标定图像示例
Fig. 4. Examples of images used in simulated calibration
下载图片 查看原文
图 5. 测距实验装置及原理示意。(a)实验平台;(b)相对距离
Fig. 5. Experimental device and principle diagram of distance measurement. (a) Experiment platform; (b) relative distance
下载图片 查看原文
图 6. 角点重建结果。(a)本文结果;(b)参考文献[
17]的结果
Fig. 6. Reconstruction results of corners. (a) Our result; (b) result from Ref. [17]
下载图片 查看原文
表 1多焦距聚焦型光场相机的仿真参数
Table1. Simulation parameters of multi-focus plenoptic camera
Parameter | /mm | b /mm | B /mm | /μm | /μm | /pixel | /mm |
---|
Value | 100 | 103.32 | 1.32 | 5 | 5 | 17 | 1.62, 1.92, 2.35 |
|
查看原文
表 2仿真数据集上的角点检测结果
Table2. Corner detection results on simulated datasets
Detection method | Recall /% | Precision /% | Mean /pixel | Std. /pixel | Time /s |
---|
Method in Ref. [16] | 14.81 | 85.31 | 1.302 | 0.598 | 205920.73 | Method in Ref. [17] | 30.96 | 39.21 | 0.822 | 0.404 | 1232.90 | Proposed method | 59.47 | 99.26 | 0.276 | 0.218 | 958.44 |
|
查看原文
表 3仿真数据集的标定结果
Table3. Calibration results on simulated datasets
Calibration method | fx | fy | K1 | K2 | cu | cv | MRE /pixel |
---|
Method in Ref.[14] | 19157.44 | 19158.12 | -2.2119 | 7785.69 | 1476.69 | 982.36 | - | Method in Ref.[16] | 19095.37 | 19108.32 | -2.1125 | 7789.89 | 1513.71 | 850.76 | 0.6609 | Method in Ref.[17] | 18979.45 | 18953.44 | -2.4074 | 8083.23 | 1507.34 | 937.27 | 0.2325 | Ours | 19018.93 | 19028.72 | -2.5763 | 8208.05 | 1485.49 | 976.05 | 0.1245 | GT | 19002.02 | 19002.02 | -2.5265 | 8170.16 | 1500.00 | 1000.00 | - |
|
查看原文
表 4R29真实数据集上的标定结果
Table4. Calibration results on R29 real dataset
Calibration method | fx | fy | K1 | K2 | cu | cv |
---|
Method in Ref.[14] | 18513.01 | 18409.14 | -2.8995 | 8728.17 | 3341.39 | 2183.02 | Method in Ref.[16] | 18609.03 | 18507.01 | -2.6519 | 8592.60 | 3349.40 | 2229.08 | Method in Ref.[17] | 18513.01 | 18409.14 | -2.8995 | 8728.17 | 3341.39 | 2183.02 | Ours | 18475.13 | 18374.66 | -2.3976 | 7946.74 | 3377.06 | 2230.79 |
|
查看原文
表 5距离测试结果
Table5. Results of distance measurement
Detection method | Calibration method | /mm | /mm |
---|
Method in Ref. [17] | Method in Ref. [14] | 3.45 | 13.98 | Method in Ref. [16] | 9.75 | 14.15 | Method in Ref. [17] | 2.35 | 14.85 | Ours | 3.12 | 14.52 | Proposed method | Method in Ref. [14] | 3.85 | 3.68 | Method in Ref. [16] | 8.48 | 3.84 | Method in Ref. [17] | 4.35 | 3.58 | Ours | 1.24 | 3.36 |
|
查看原文
刘青松, 谢晓方, 张烜喆, 田瑜, 李俊, 王彦, 许晓军. 用于聚焦型光场相机标定的棋盘角点检测算法[J]. 光学学报, 2020, 40(14): 1415002. Qingsong Liu, Xiaofang Xie, Xuanzhe Zhang, Yu Tian, Jun Li, Yan Wang, Xiaojun Xu. Checkerboard Corner Detection Algorithm for Calibration of Focused Plenoptic Camera[J]. Acta Optica Sinica, 2020, 40(14): 1415002.