LEGENDFTW/image-filtering-explorer

🔍 Image Filtering & Denoising Explorer

An interactive Streamlit app for exploring spatial image filters and their denoising behaviour. Built as a course assignment for an image analysis module.

What it does

• —Upload your own image or choose from 3 built-in noisy samples
• —Select one of 5 classic filters: Gaussian Blur, Median, Bilateral, Box (Mean), Non-local Means
• —Tune filter parameters with live sliders
• —View side-by-side input/output and a difference image
• —Get quantitative metrics: PSNR, SSIM, noise reduction estimate
• —Explore histogram comparison and a local noise heatmap
• —Add synthetic noise (Gaussian / Salt & Pepper / Speckle) for controlled experiments
• —Read concise theory explanations for each filter

Run locally

\\\bash git clone <your-repo-url> cd image-filtering-app pip install -r requirements.txt streamlit run app.py \\\

Hugging Face Space

🔗 Live demo: (add your URL here after deployment)

Known limitations

• —Non-local Means can be slow (5–15s) on images larger than ~512×512px
• —Metrics compare input vs output — without a true clean reference, PSNR/SSIM measure change, not absolute quality
• —Very large uploads (>800px) are auto-resized