Image Median Filter Tool
Professionally remove salt-and-pepper noise, speckles, and dust from your photographs while perfectly preserving sharp edges. Processed securely within your browser.
Select an Image
Supports JPG, PNG, WEBP
Max processing size: 1000pxLive Render Preview
Upload a noisy photograph to
apply the pixel sorting filter.
What is an Image Median Filter?
The Image Median Filter Tool is a professional-grade, browser-based digital image processing utility. In digital signal processing, a median filter is a non-linear technique predominantly used to remove "salt and pepper" noise (random black and white speckles) or dust scratches from a photograph or scanned document.
How Does the Median Sorting Algorithm Work?
Unlike standard blur filters (like Gaussian Blur) that mathematically average out pixel values—which inadvertently causes sharp edges to become fuzzy and out-of-focus—a median filter operates through spatial pixel sorting:
- Grid Selection: The algorithm looks at a "window" of pixels surrounding every single pixel in your image. For example, a radius of 1 creates a 3x3 grid (9 pixels). A radius of 3 creates a 7x7 grid (49 pixels).
- Value Gathering: It collects the color values (Red, Green, and Blue) of all pixels within this specific window.
- Numerical Sorting: It sorts these values in strict numerical order and selects the exact middle (median) value.
- Pixel Replacement: It replaces the original center pixel's color with this new median value.
Because the filter chooses an existing median value from the neighborhood rather than blending colors together, sharp edges (like the boundary between a black object and a white background) are perfectly preserved, while rogue speckles of noise are entirely erased.
Median Filter vs. Gaussian Blur
If you have an old, grainy photograph or a scanned text document with dust marks, applying a Gaussian Blur will simply smudge the dust marks into fuzzy gray blobs, destroying the readability of the underlying text or subject. A Median Filter, however, recognizes that the dust mark is a statistical outlier compared to the surrounding pixels. It completely eliminates the dust mark while leaving the text crisp, clean, and legible. It is the absolute best tool for restoring scanned archival documents and distressed vintage photography.
Step-by-Step Guide: How to Clean a Noisy Image
- Upload Your Image: Drag and drop your noisy JPG, PNG, or WEBP file into the secure upload zone.
- Select the Filter Radius: Use the slider to determine the size of the pixel grid. A 1px radius (3x3 grid) is perfect for light speckles. A 3px radius (7x7 grid) is better for heavy, clustered noise.
- Apply the Filter: Click the "Apply Filter" button. The browser will begin sorting millions of pixels. Note: High radii on large images may take 3-10 seconds to process.
- Review and Download: Evaluate the live preview to ensure the noise is gone without losing too much structural detail. Click "Download Cleaned Image" to save the high-resolution PNG to your device.
100% Privacy & Client-Side Security
Protecting your personal and archival data is our top priority. Traditional photo-editing websites force you to upload your sensitive images to remote cloud servers. Our architecture completely bypasses the cloud. All complex image sorting is executed locally by your device's browser using HTML5 Canvas. Your photos never leave your computer, ensuring absolute privacy and zero network latency.
Frequently Asked Questions (FAQ)
Why does a high filter radius take so long to process?
Median filtering is incredibly computationally intensive. If you have an 800x800 image and set a radius of 4 (a 9x9 grid), your browser's processor has to extract, sort, and select the median from an array of 81 numbers, 640,000 times, for every color channel. We cap image sizes to prevent your browser from freezing, but higher numbers will naturally require a few seconds of loading.
Why did my image lose small details like textures or leaves?
While median filters preserve sharp macro-edges (like the edge of a building), they are specifically designed to eliminate small, isolated clusters of pixels. If a texture (like the pores on skin or leaves on a distant tree) is smaller than the filter's grid size, the algorithm will classify it as "noise" and remove it. Always use the lowest radius necessary to clear the noise.
Do I own the commercial rights to the cleaned image?
Yes. Because the tool operates purely as a mathematical filter on your own machine, you retain full ownership and copyright of any output generated, provided you hold the rights to the original photograph you uploaded.