/Project Details
Classical Image Processing in MATLAB
A collection of MATLAB image processing projects covering spatial enhancement, image restoration, frequency-domain filtering, segmentation, and morphology.
This project groups three classical image processing studies into one coherent portfolio entry. The focus is not deep learning; it is the lower-level machinery behind image enhancement, restoration, segmentation, and morphology.
The spatial enhancement module implements manual histogram computation, threshold-based binary enhancement, median filtering for salt-and-pepper noise, Sobel edge comparison, and min/max filtering with explicit padding logic.
The restoration module moves between spatial and frequency domains. It uses midpoint filtering for uniform noise and FFT-based Butterworth notch reject filtering for periodic noise, then compares edge maps before and after reconstruction.
The segmentation module detects dice and counts pips using grayscale conversion, Otsu thresholding, morphological closing/opening, connected component labeling, and object masks.
Highlights
- Structured the three studies as a coherent image-processing portfolio entry spanning enhancement, restoration, and segmentation.
- Implemented histogram thresholding and binary enhancement without relying on high-level shortcuts.
- Used median, midpoint, min/max, Sobel, and Butterworth notch reject filters across spatial and frequency domains.
- Built a dice segmentation pipeline with Otsu thresholding, morphology, connected components, and pip counting.
- Showed classical computer-vision reasoning through explicit filtering, thresholding, morphology, and frequency-domain operations.
Figures
