Techniques for the segmentation of striation patterns

Area of Science:

• Image Processing
• Forensic Science
• Pattern Recognition

Background:

• Segmentation of faint striation patterns from isotropic backgrounds is challenging.
• Existing methods for local anisotropy determination often require single images.
• Faint tool marks in forensic science present difficulties in robust segmentation.

Purpose of the Study:

• To develop and validate an image processing technique for robust segmentation of faint striation patterns.
• To present an alternative strategy using an illumination series instead of single images.
• To demonstrate the effectiveness of the proposed method on forensic tool mark segmentation.

Main Methods:

• Utilizing an illumination series obtained by systematically varying the azimuth of spot illumination.
• Analyzing the local contrast characteristics with respect to the illumination azimuth.
• Exploiting the property that striations show pronounced contrast maxima under perpendicular illumination, unlike isotropic backgrounds.

Main Results:

• The proposed technique successfully segments faint striation patterns from isotropic backgrounds.
• Experimental results confirm the methodology's ability to correctly segment faint tool marks.
• The approach demonstrates superior performance compared to methods relying on single images.

Conclusions:

• The illumination series technique provides a robust method for segmenting faint striation patterns.
• This image processing approach enhances the analysis of forensic evidence, specifically tool marks.
• The findings contribute to improved accuracy and reliability in forensic image analysis.