Contrast Enhancement

Enhance image contrast by normalizing the histogram to use the full dynamic range.

Enhance image contrast using histogram normalization (the algorithm from GIMP's Auto Levels). This block stretches the image histogram to use the full available range [0-255], improving visibility of features with low contrast.

How This Block Works

1. Channel Analysis: For grayscale images, find min/max directly. For color images, analyze each channel independently.

2. Histogram Normalization: For each channel, stretch values from [min, max] to [0, 255] using linear scaling: output = (input - min) / (max - min) * 255

3. Clipping: Values outside [0, 255] are clipped to the valid range

Common Use Cases

• Low-contrast medical imaging: Normalize tissue visibility across varying acquisition parameters

• Industrial inspection: Enhance subtle surface defects on dull materials

• Surveillance footage: Improve nighttime or backlit scene visibility

• Document scanning: Brighten poorly lit document photos

• Microscopy: Boost signal from weak fluorescence or phase-contrast images

Input Parameters

image : Input image to enhance (color or grayscale)

• Can be single-channel, 3-channel (BGR), or 4-channel (BGRA)

• Each channel is normalized independently for color images

clip_limit : Percentage of histogram range to skip at extremes (default: 0)

• Range: 0-50

• 0: No clipping, entire histogram from min to max is used

• 1-3: Skip dark and bright outliers (robust to noise)

• 5-10: Very aggressive outlier removal

• 20-50: Extreme outlier removal, may lose subtle details

contrast_multiplier : Multiplier for contrast scaling after normalization (default: 1.0)

• Range: 0.1-5.0

• 1.0: No additional scaling, just normalization

• 0.5-0.9: Reduce contrast for smoother images

• 1.1-2.0: Increase contrast for more dramatic enhancement

normalize_brightness : Apply brightness normalization using midtone equalization (default: False)

• False: Only histogram normalization and contrast scaling

• True: After histogram normalization, apply midtone adjustment for balanced brightness

Outputs

image : Enhanced image with normalized contrast, same shape and type as input

Notes

• Sensitive to outliers: Extreme min/max values (single dark/bright pixels) stretch most of the histogram into a narrow range. Use morphological opening (Morphological Transformation v2) as preprocessing to remove spurious dark/bright specks.

• Color shift: For color images, each channel stretches independently, which can shift hue if channels have very different dynamic ranges

• Efficiency: Very fast - linear scan for min/max, then linear transformation per pixel

• Brightness normalization: When enabled, applies midtone stretch (gamma ≈ 1.3) for more balanced perceived brightness

Type identifier

Use the following identifier in step "type" field: roboflow_core/contrast_enhancement@v1 to add the block as a step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
clip_limit	int	Percentage of histogram range to skip at dark and bright extremes. 0: use full range from min to max. 1-3: skip outliers (robust). 5-10: very aggressive outlier removal..	❌
contrast_multiplier	float	Multiplier for contrast scaling after normalization. 1.0: no additional scaling (just histogram normalization). <1.0: reduce contrast. >1.0: increase contrast for more dramatic enhancement..	❌
normalize_brightness	bool	Apply brightness normalization using midtone equalization. When False, only histogram normalization and contrast scaling are applied. When True, applies midtone adjustment for more balanced perceived brightness..	❌

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Contrast Enhancement in version v1 has.

Input and output bindings

• input

  • image (image): Image to enhance (color or grayscale). Each channel is normalized independently..

• output

  • image (image): Image in workflows.

Example JSON definition

{
	    "name": "<your_step_name_here>",
	    "type": "roboflow_core/contrast_enhancement@v1",
	    "image": "$inputs.image",
	    "clip_limit": "<block_does_not_provide_example>",
	    "contrast_multiplier": "<block_does_not_provide_example>",
	    "normalize_brightness": "<block_does_not_provide_example>"
	}