Improvement of the amplitude image through statistical filtering

Area of Science:

• Image Processing
• Signal Analysis
• Data Science

Background:

• Automatic thresholding of amplitude images is crucial for accurate phase analysis.
• Existing methods may lack precision or require manual parameter tuning.
• Probability density-based approaches offer a robust alternative for image segmentation.

Purpose of the Study:

• To introduce a novel filter for automatic thresholding of amplitude images derived from phase analysis.
• To establish a method for generating significant amplitude images by effectively zeroing noise.
• To provide a statistically grounded filter applicable to various signal processing scenarios.

Main Methods:

• Development of a filter based on the probability density of signal amplitude.
• Implementation of a thresholding criterion: amplitude squared < 4a0 log(1/alpha)/N.
• Analysis of the statistical power of the proposed detection test.

Main Results:

• The proposed filter successfully generates images of significant amplitude by setting non-significant values to zero.
• The thresholding formula is defined in terms of signal mean (a0), error risk (alpha), and frame count (N).
• The study discusses the statistical power of the test and offers modifications for pre-processed data.

Conclusions:

• The probability density-based filter provides an effective automated solution for amplitude image thresholding in phase analysis.
• The method enhances the identification of significant amplitude features, improving data interpretation.
• The filter's adaptability for pre-processed data broadens its applicability in scientific imaging.