This repository contains a high-performance Red Color Segmentation system implemented using OpenCV on Raspberry Pi 4B. By utilizing dual-range HSV thresholding and morphological refinement, this project achieves high selectivity in isolating red objects across diverse environments.
- Dual-Range HSV Thresholding: Implements a dual-masking technique to capture both ends of the Red Hue spectrum (0-10 and 170-180).
- Morphological Refinement: Uses Opening and Closing operations to eliminate salt-and-pepper noise and bridge structural gaps.
- Lighting Robustness: Tested under various light factors (0.4x to 1.6x Value) to ensure stability in different environments.
- Multi-Domain Validation: Proven effective on Automotive, Horticulture (Apples), and Botany (Strawberries) datasets.
| Scenario | Automotive | Apples | Strawberries |
|---|---|---|---|
| Default | 9.16% | 41.88% | 11.85% |
| Tight | 3.77% | 3.60% | 7.96% |
| Loose | 10.26% | 54.69% | 12.81% |
Below is the vertical showcase of the segmentation pipeline results.
Original Input Image
Binary Mask
Final Segmentation Result
Original Input Image
Binary Mask
Final Segmentation Result
Original Input Image
Binary Mask
Final Segmentation Result
Click the thumbnail below to watch the technical walkthrough and live simulation:
Explore the full technical analysis and presentation slides:
- 📑 Technical Report (PDF) - Detailed methodology, theory, and experimental data.
- 📊 Project Presentation (PPTX) - Visual summary of the project pipeline and results.
Author: Ahmad Hanif Abiyyu Khrisna
Institution: Electronic Engineering Polytechnic Institute of Surabaya (PENS)