How To: Assess Image Quality Before Pipeline Design

Run diagnostics on a plate image to objectively assess noise, contrast, and structure before choosing enhancers and detectors.

from phenotypic.data import load_yeast_plate
from phenotypic.plotting import PlotDiagnostics
from phenotypic.util import ImageMetricsCalculator
import matplotlib.pyplot as plt

plate = load_yeast_plate()
fig = PlotDiagnostics().inspect(plate)
fig

Decision Guide

Metric	Threshold	Action
Low SNR (< 10)	Noisy image	Add StableDenoise or BlurGauss
Low RMS contrast	Faint colonies	Add CLAHE or ContrastStretching
Low dynamic range	Under-exposed	Add ContrastStretching
Low gradient mean	Soft edges	Add UnsharpMask or FocusEdgeSobel
Long correlation length	Uneven illumination	Add HomomorphicFilter

calculator = ImageMetricsCalculator(plate.detect_mat[:])
metrics = {
    "noise": calculator.compute_noise_metrics(),
    "contrast": calculator.compute_contrast_metrics(),
    "structure": calculator.compute_structure_metrics(),
    "background": calculator.compute_background_metrics(),
}
for category, values in metrics.items():
    print(f"\n{category}:")
    for key, val in values.items():
        if isinstance(val, (int, float)):
            print(f"  {key}: {val:.4f}")

noise:
  snr: 16.8730
  sigma_mad: 0.0197
  correlation_length: 49.5000

contrast:
  rms_contrast: 0.2567
  michelson: 0.4287
  dynamic_range: 0.0022
  p1: 0.2520
  p99: 0.6302

structure:
  mean_coherence: 0.2913
  optimal_scale: 1.0000
  peak_response: 0.0756

background:
  nonuniformity_ratio: 0.1346
  mean_gradient: 0.0005

plt.close("all")