Author Affiliations
1Key Laboratory of Advanced Manufacturing and Intelligent Technology, Ministry of Education, Harbin 150080, China2Institute of Mechanical & Power Engineering, Harbin University of Science and Technology, Harbin 150080, Chinashow less
Fig. 1. 朗伯光学反射模型Lambertian optical reflection model
Fig. 2. 瓷水杯在均光处理前后的图片对照, 取样过程和像素变化Picture comparison, sampling process and pixel change of porcelain cup before and after homogenizing treatment
Fig. 3. 传统均值漂移算法示意图Schematic diagram of traditional mean shift algorithm
Fig. 4. 传统均值漂移算法缺陷Traditional mean shiftalgorithm defect
Fig. 5. 激光点投射变化情况Laser spot projection changes
Fig. 6. 激光点群对均值漂移的算法影响Influence of laser point group on mean shift algorithm
Fig. 7. 漂移描述子功能示意图Drift description subfunction diagram
Fig. 8. 深度图平滑处理前后比较示意图Schematic diagram of before and after comparison of depth map smoothing
Fig. 9. Tsukuba图片的原图、深度图、重建图比较Comparison of original images, disparity maps and reconstruction maps of Tsukuba images
Fig. 10. 实拍图片重建效果Real shot image reconstruction effect
Fig. 11. 显示器背面的尺寸取样示意图Sampling of dimensions on the back of the monitor
Fig. 12. 显示器重建前后的样本尺寸值Sample size values before and after monitor reconstruction
Fig. 13. 各方法RMSE性能评价RMSE performance evaluation of each method
Fig. 14. 各方法PSNR和SSIM性能评价PSNR and SSIM performance evaluation of each method
Direction | Top left | Top right | Left | Central | Right | Bottom | Bottom right | τ | 118.60 | 90.53 | 106.0 | 96.26 | 29.22 | 110.01 | 67.04 | δ | 89.38 | 86.52 | 86.52 | 75.37 | 85.37 | 75.06 | 96.83 |
|
Table 1. Optical parameter azimuth and inclination estimation (Unit: degree)
The name of algorithm | Image size/pixel | Time consuming/ms | GC algorithm | 384×288 | 2 272 | Ref. [10] algorithm | 384×288 | 1 702 | Algorithm of this paper | 384×288 | 1 878 |
|
Table 2. Time-consuming reference for standard test image reconstruction experiment algorithm under different methods