• Acta Optica Sinica
  • Vol. 40, Issue 5, 0520001 (2020)
Lin Fu1, Haibo Hong2, Xi Wang1, Gaobo Xiao1, and Mingjun Ren1、*
Author Affiliations
  • 1State Key Laboratory of Mechanical System and Vibration, School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
  • 2Shanghai Spaceflight Precision Machinery Institute, Shanghai 200240, China
  • show less
    DOI: 10.3788/AOS202040.0520001 Cite this Article Set citation alerts
    Lin Fu, Haibo Hong, Xi Wang, Gaobo Xiao, Mingjun Ren. Non-Lambertian Photometric Stereo Vision Based on Inverse Reflectance Model[J]. Acta Optica Sinica, 2020, 40(5): 0520001 Copy Citation Text show less
    Ray angle diagram
    Fig. 1. Ray angle diagram
    Relationship between nTv and observation of different materials. (a) Blue fabric; (b) brass; (c) specular white phenolic; (d) black obsidian
    Fig. 2. Relationship between nTv and observation of different materials. (a) Blue fabric; (b) brass; (c) specular white phenolic; (d) black obsidian
    Method framework
    Fig. 3. Method framework
    Effects of elevation angle on the following factors. (a) Relative error of nTl; (b) overlapped region ratio; (c) condition number; (d) angular error of surface normal
    Fig. 4. Effects of elevation angle on the following factors. (a) Relative error of nTl; (b) overlapped region ratio; (c) condition number; (d) angular error of surface normal
    Rendered images of rabbit and ball model
    Fig. 5. Rendered images of rabbit and ball model
    Comparison of normal vector recovery maps of four materials (from top to bottom: pvc, alum-bronze, ipswich-pine-221, color-changing-paint2)
    Fig. 6. Comparison of normal vector recovery maps of four materials (from top to bottom: pvc, alum-bronze, ipswich-pine-221, color-changing-paint2)
    Comparison of four material normal vector angular errors (from top to bottom: pvc, alum-bronze, ipswich-pine-221, color-changing-paint2)
    Fig. 7. Comparison of four material normal vector angular errors (from top to bottom: pvc, alum-bronze, ipswich-pine-221, color-changing-paint2)
    Angular error of synthetic surface normal vector based on MERL dataset
    Fig. 8. Angular error of synthetic surface normal vector based on MERL dataset
    Device concept map
    Fig. 9. Device concept map
    Images of the copper bullet under four lights
    Fig. 10. Images of the copper bullet under four lights
    Bullet recovery normal map and bump map. (a) Recovery map of surface normal; (b) bump map
    Fig. 11. Bullet recovery normal map and bump map. (a) Recovery map of surface normal; (b) bump map
    Dataset materialBallRabbit
    LSCBROursLSCBROurs
    Pvc24.62.120.52126.72.290.464
    Alum-bronze28.63.930.15331.94.260.172
    Ipswich-pine-22112.91.850.22114.51.730.262
    Color-changing-paint29.891.510.61311.61.580.573
    Mean angular error12.72.130.37513.82.170.365
    Table 1. Angular errors of simulation(°)
    MethodLSCBROurs
    Angular error /(°)31.628.46.91
    Computation time /s2.1100.65
    Table 2. Angular errors and computation time of real experiments
    Lin Fu, Haibo Hong, Xi Wang, Gaobo Xiao, Mingjun Ren. Non-Lambertian Photometric Stereo Vision Based on Inverse Reflectance Model[J]. Acta Optica Sinica, 2020, 40(5): 0520001
    Download Citation