Automatic estimation of a scale resolution in forensic images.

Area of Science:

• Forensic Science
• Image Analysis
• Computer Vision

Background:

• Accurate scale resolution is crucial for forensic image analysis.
• Manual scale detection can be time-consuming and prone to errors.
• Digital comparison of forensic images requires consistent scaling.

Purpose of the Study:

• To propose and evaluate a novel method for automatic scale resolution detection in shoeprint images.
• To enable automatic rescaling of forensic images for consistent analysis.
• To facilitate digital comparison of shoeprint evidence.

Main Methods:

• A method creating a vector of correlations from co-occurrence matrices for each image row.
• Scale resolution estimation using maxima in the Fourier spectrum of correlation vectors.
• Evaluation on over 500 forensic and laboratory-acquired shoeprint images.

Main Results:

• The proposed method successfully estimates scale resolution in shoeprint images.
• Experimental results demonstrate the feasibility of automatic scale detection.
• The technique is effective on images from both crime scenes and forensic labs.

Conclusions:

• The automatic detection of scale resolution is achievable with the proposed method.
• This technique can automate the rescaling of forensic images for 'one-to-one' printing.
• Automated rescaling facilitates digital comparison of shoeprint evidence.