Estimation of the probability density function of random displacements from images

Area of Science:

• Image analysis
• Statistical physics
• Computational imaging

Background:

• Conventional methods estimate particle displacement probability density functions (PDFs) using cross-correlation (CC), assuming Gaussian distributions for particle profiles and displacements.
• These assumptions limit the accuracy and applicability of CC methods in complex scenarios.

Purpose of the Study:

• To introduce a novel image-based algorithm, image-based probability estimation of displacement (iPED), for determining particle displacement PDFs.
• To overcome the limitations of conventional CC methods by removing assumptions about particle intensity profiles and displacement distributions.

Main Methods:

• Developed an image-based algorithm (iPED) to estimate the probability density function (PDF) of particle displacements directly from image sequences.
• Incorporated a statistical convergence criterion within iPED for accurate PDF estimation.
• Compared iPED performance against traditional cross-correlation (CC) methods using both synthetic and real experimental image data.

Main Results:

• iPED accurately resolves the PDF of particle displacements without making underlying assumptions about particle shape or displacement distributions.
• The method demonstrates robust performance across both Gaussian and non-Gaussian particle profiles and displacement processes.
• Validation using synthetic images confirmed iPED's accuracy, and application to real experimental data showcased its practical utility.

Conclusions:

• iPED offers a generalized, assumption-free approach for estimating the probability density function of random displacements from images.
• This image-based method is applicable to diverse moving objects of arbitrary shapes and various underlying physical processes.
• iPED significantly advances particle displacement analysis in fields relying on image-based measurements.