BreadScan — Bread Crumb Porosity Analyzer

BreadScan Literature-Grounded · Multi-Parameter Fingerprint

Thresholding: Gonzales-Barron & Butler (2006) J.Food Eng.
Brown bread: Bosakova-Ardenska (2015) HisAnalysis peak-mode
Fingerprint: Ruderman et al. (2025) ScienceDirect

Image

🍞

Drop or click to upload
Cross-section photo or flatbed scan
ROI: crumb only, no crust edges

Bread Type

Algorithm: Sauvola local adaptive Computes a separate threshold for each pixel based on local mean and standard deviation. Handles uneven lighting from photographic images. Applied on CLAHE-enhanced grayscale.

ROI (Region of Interest)

ROI side (mm)40

Advanced — working resolution

Working resolution (px/mm)10

Constant working resolution makes Sauvola's window and the closing kernel correspond to a fixed physical size, eliminating scale-driven variability between repeats.

Free rectangle — draw any ROI to fit the loaf cross-section. Fixed square samples a deterministic *N*×*N* mm window for strict reproducibility benchmarks.

Preprocessing

Gaussian σ (denoising)1.0

CLAHE clip (contrast)2.0

Why CLAHE first? Enhances local contrast before thresholding. Used by Lassoued et al. (2008) for heterogeneous crumb structures.

BSG Mode — Threshold Offset

Peak offset (±% of range)+5%

HisAnalysis logic: threshold = histogram peak + offset, computed on RAW grayscale (no CLAHE). Negative offset (default −8%) = detect only pixels significantly darker than the dominant crumb color. More negative = more pores. Tune until binary view matches visible pores.

Min cell area20

Max cell area15000

No circularity filter by default. Literature shows bread pores are elliptical (circularity < 1), not circular — filtering by shape rejects real pores. Gonzales-Barron & Butler 2006.

Output

3D correction ×1.50

Session Results

No analyses saved yet.

Each successful analysis is appended to this browser's local store. Survives reload; private to this device.

Void Fraction

—

2D porosity %

3D Corrected

—

estimated vol. %

Cell Count

—

valid pores

Cell Density

—

cells/cm² (crumb fineness)

Mean Cell Area

—

px² per pore

Digital Texture Fingerprint after Ruderman et al. (2025) Bread Texture Analyser · ScienceDirect

Void Fraction

—

% crumb area as pores

Cell Density

—

cells/100px² (fineness)

Mean Cell Area

—

px² (coarseness)

Cell Uniformity

—

1 − CV (1=uniform)

Wall Thickness

—

px (estimated)

Macro Pore %

—

% large cells (>2000px²)

Original

Upload

CLAHE enhancedLassoued 2008

—

Binary maskSauvola / HisAnalysis

—

Colormap by areaRuderman 2025

—

Histogram peaksBosakova 2015

—

Cell Size Distribution — Zghal et al. (1999) parameters —

Run analysis

Analysis Log

No analysis run yet.

What's in your photo?

Calibrate with ruler