Developed a computer vision pipeline for detecting and segmenting characters on vehicle license plates. The project emphasizes classical image processing techniques for robust feature extraction and segmentation.
Technical Features
Image Preprocessing
Applied grayscale conversion, noise reduction, and normalization to prepare images for analysis.
Used thresholding and adaptive binarization for robust character segmentation under varying lighting conditions.
Feature Extraction & Segmentation
Implemented morphological operations (erosion, dilation, opening, closing) to enhance character structures.
Applied contour analysis to detect and isolate individual characters.
Filtered candidate regions using size, aspect ratio, and geometric constraints for accurate detection.
Pipeline Integration
Developed a modular pipeline combining preprocessing, segmentation, and filtering for end-to-end detection.
Designed for batch processing and scalability to handle multiple license plate images efficiently.