Decomposition of shading and reflectance from a texture image

Area of Science:

• Computer Vision
• Image Processing
• Computational Photography

Background:

• Separating intrinsic image properties like shading and reflectance is crucial for image analysis and manipulation.
• Existing methods often struggle with complex textures and varying illumination conditions.

Purpose of the Study:

• To present a robust method for decomposing shading and reflectance from color texture images.
• To enhance the accuracy of reflectance estimation by addressing shading variations.

Main Methods:

• A novel approach computes or synthesizes shading intensity for each pixel based on its characteristics.
• Reflectance is subsequently decomposed using the estimated shading information.
• An optional denoising step is incorporated to refine the decomposed components.

Main Results:

• The proposed method successfully decomposes shading and reflectance components from real color texture images.
• Evaluation demonstrates the effectiveness of the shading computation/synthesis and subsequent reflectance decomposition.
• The method shows potential for improving image analysis tasks reliant on intrinsic image properties.

Conclusions:

• The presented method offers an effective solution for separating shading and reflectance in color textures.
• Accurate decomposition facilitates further analysis and manipulation of intrinsic image properties.
• The technique shows promise for applications in computer vision and image editing.