Edge Modeling by Two Blur Parameters in Varying Contrasts.

Area of Science:

• Image processing and computer vision
• Computational imaging and optics

Background:

• Accurate edge profile modeling is crucial for image analysis.
• Existing models often simplify edge characteristics, limiting their applicability.
• Varying illumination and camera-object distances complicate edge parameter estimation.

Purpose of the Study:

• To develop and validate a two-blur parameter edge model.
• To establish a method for estimating and predicting edge parameters under diverse conditions.
• To compare the proposed model's performance against a one-blur parameter model.

Main Methods:

• Mathematical and experimental validation of the two-blur parameter edge model.
• Brute-force estimation of blur parameters for observed edge profiles.
• Kriging (surface interpolation) for predicting edge parameters with varying brightness.
• Root mean squared error analysis for model comparison.

Main Results:

• The proposed two-blur parameter edge model demonstrates mathematical and experimental validity.
• Edge parameters exhibit a global trend dependent on dark-side and light-side brightness, consistent across varying camera-to-object distances.
• The two-blur parameter model shows superior performance compared to a one-blur parameter model in most tested scenarios.

Conclusions:

• The developed two-blur parameter edge model provides a more comprehensive representation of image edges.
• The prediction method effectively captures the relationship between edge parameters and brightness variations.
• This enhanced modeling approach offers improved accuracy for image analysis tasks involving complex edge profiles.