 |
 |
| Target | Result |
|
|
|
| Target | Result |
| We developed a method to estimate 3D shape of opaque object. Input images with different light source direction is used. This method can estimate the shape under unknown light source directions. We use the constraints that the light intensity is constant and the shape at the object boundary is known. The ambiguity problem of uncalibrated photometric stereo is solved using these constraints, which results in unique shape. We also developed a method to improve the shape using polarization. |