A high-performance document scanning application built with Python and OpenCV. This project features a custom-engineered Hybrid Edge Detection Pipeline designed to isolate documents from complex backgrounds with high precision.
- Custom Sobel Implementation: Unlike standard solutions, the core edge detection (Sobel Filter) is implemented from scratch using NumPy to handle mathematical convolutions and image gradients manually.
- Hybrid Detection Pipeline: Combines Sobel Gradients with HSV Color Masking to filter out background noise and focus strictly on the document area.
- Perspective Transformation: Automatically detects the four corners of a document and applies a 4-point transform to generate a top-down, "scanned" view.
- Adaptive Processing: Utilizes Fast Non-Local Means Denoising and Adaptive Thresholding to produce crisp, high-contrast outputs suitable for printing.
- Advanced GUI: A modern, dark-themed dashboard built with Tkinter, featuring a multi-threaded architecture to ensure smooth UI performance during processing.
- Language: Python
- Libraries: OpenCV, NumPy, Pillow (PIL)
- UI: Tkinter (Custom Dark Theme)
- Algorithms: Sobel Filter (Manual), Perspective Warp, Adaptive Gaussian Thresholding.
The application includes a Pipeline Viewer that allows you to see the image at every stage of transformation:
- Grayscale -> 2. Gaussian Blur -> 3. Sobel (From Scratch) -> 4. HSV Masking -> 5. Hybrid Combination -> 6. Contour Detection -> 7. Perspective Warp -> 8. Final Adaptive Threshold.
- Clone the repo: git clone https://github.com/MariamAshraf25/Smart-Document-Scanner.git
Mariam Ashraf Computer Engineering Student - Faculty of Engineering - Capital University (Formerly Helwan) LinkedIn Profile