Accurate image quantization adapted to multisource photometric reconstruction for rough textured surface analysis.

Area of Science:

• Computer Vision
• 3D Reconstruction
• Photometric Stereo

Background:

• Classical photometric stereo (PS) reconstructs 3D surfaces from images under varying light. Real-world images are limited and noisy, challenging traditional PS accuracy.
• Accurate 3D surface reconstruction is crucial for applications in robotics, augmented reality, and digital archiving.

Purpose of the Study:

• To present an accurate 3D recovery approach for real textured surfaces using an improved photometric stereo method.
• To address the limitations of traditional PS in handling noisy and quantized real-world image data.

Main Methods:

• The proposed method utilizes a sequence of images acquired under each light source.
• This approach aims to recover an accurate and unlimited representation of the surface's 3D geometry.
• Performance is evaluated by comparing the proposed method against traditional PS techniques on real textured surfaces.

Main Results:

• The developed photometric stereo method demonstrates enhanced accuracy in 3D recovery for real textured surfaces.
• The use of image sequences per light source effectively mitigates noise and quantization issues inherent in real-world imaging.
• Comparative analysis validates the superior performance of the proposed method over traditional PS techniques.

Conclusions:

• The proposed photometric stereo approach offers a significant advancement for accurate 3D surface reconstruction from real-world data.
• This method provides a more robust and detailed 3D representation compared to conventional techniques.
• The findings contribute to improving the reliability of 3D reconstruction in practical, unidealized scenarios.