Retinex / Illumination Correction
Correct uneven lighting and recover shadow detail with Retinex-based illumination correction for OCR prep, inspection, and image readability workflows.
Retinex illumination correction is useful when an image suffers from glare, dim corners, harsh shadow gradients, or lighting that makes later analysis harder. It helps normalize illumination before OCR, thresholding, segmentation, inspection, or visual review by lifting dark regions without flattening all local detail.
How to Use Retinex / Illumination Correction
- Upload an image: Upload one of the formats listed by the tool (JPG, PNG, WebP) and wait for the local preview.
- Adjust settings and process: Choose multiscale or single-scale correction, then tune illumination strength, color restoration, and detail boost.
- Review and download: Review the result, then download it in one of the formats offered by this tool.
Controls and Options
Raise illumination strength first, then fine-tune color restoration and detail boost until shadows lift without flattening texture.
Formats and Output
Input: JPG, PNG, WebP. Output: PNG, JPG, or WebP. Retinex correction lifts uneven illumination, recovers shadow detail, and preserves useful local contrast for analysis.
Why Retinex Illumination Correction Matters
Retinex correction is useful when an image contains glare, dim shadows, backlit regions, or uneven lighting that makes the content harder to inspect. It is especially effective before OCR, thresholding, segmentation, and analysis tasks where illumination variation hides edges or text that should be easy to read.
How the Workflow Helps
Instead of treating the whole image as one exposure problem, Retinex-style correction separates broad illumination patterns from local texture. That makes it possible to lift dark regions, compress overly bright patches, and retain local structure so the image remains readable for later processing.
When to Use It
- OCR and document prep when receipts or scans contain bright spots and dark corners
- Inspection and QA where uneven lighting hides surface detail
- Preprocessing for thresholding when segmentation depends on stable local contrast
- Image analysis experiments that need cleaner illumination before feature extraction
Frequently asked questions
When is Retinex useful?
Retinex is useful for shadow-heavy document captures, indoor photos with uneven light, low-contrast frames, and scenes where you want more balanced illumination before further analysis.
What is multi-scale Retinex?
Multi-scale Retinex blends illumination estimates from several blur scales, which makes it more stable across both fine detail and larger lighting gradients.
What does color restoration do?
Color restoration helps maintain believable color separation after contrast and illumination normalization so the result does not look overly gray or washed out.
Should I use this before OCR or segmentation?
Yes, it can be useful as a preprocessing step when shadows or uneven lighting are reducing contrast before OCR, thresholding, or classical computer vision analysis.
Does this run locally?
Yes. Retinex processing runs directly in your browser via OpenCV WASM with no file upload.
What image formats are supported?
JPG, PNG, and WebP are supported.