Measure Entropy

Quantifies the amount of change or visual complexity in an image using entropy-based measurements.

Remarks

The Measure Entropy command analyzes an image to determine its level of detail, noise, or "busyness" by calculating entropy or related metrics. This is useful for assessing image quality, detecting excessive noise or background patterns, and supporting automated decisions in IP Profiles or custom workflows.

How It Works

1. The image is optionally binarized or converted to grayscale, depending on configuration.
2. The command computes an entropy metric based on the selected 'Measurement Type':
   • Count: Measures the number of pixel transitions (changes) as a fraction of the maximum possible, indicating the prevalence of edges or noise.
   • Value: Measures the total magnitude of pixel changes, reflecting both the number and strength of transitions.
   • BPI: Calculates the number of blobs per square inch, useful for quantifying isolated marks or specks.
3. The result is a single numeric value representing the selected entropy metric.

Supported Pixel Formats

• All common pixel formats are supported as input. Images are automatically converted to grayscale or binarized as needed for analysis.

Diagnostics

When diagnostic mode is enabled, Measure Entropy logs the computed entropy value and, if binarization is applied, generates a diagnostic image showing the binarized result. This output is useful for tuning quality control thresholds and understanding how entropy is measured.

Classification Features

The command produces a single numeric feature (entropy value) for each image. This feature can be used for:

• Automated quality control (e.g., flagging noisy or cluttered images)
• Comparing before/after results of image cleanup operations
• Supporting workflow branching or custom logic based on image complexity

Configuration Guidance

• Select the 'Measurement Type' based on the type of quality issue you want to detect.
• Enable 'Binarization' for more consistent results on documents with variable backgrounds or print quality.
• Use diagnostic output to review entropy values and refine your thresholds for flagging or classification.

Properties

See Also

Binarize

Used By

IP Step