C++ Shape Detection System

A high-performance computer vision application for detecting rectangles and spheres/obloids in images. Features rotation-invariant detection algorithms with comprehensive testing suite.

Demo

Overview

• Dual Shape Detection: Detects rectangles and spheres/obloids (circles, ellipses)
• Rotation Invariant: 100% detection accuracy across all rotation angles (0°-180°)
• High Performance: Multi-threaded processing with OpenMP parallelization
• No Dependencies: Pure C++ implementation without external CV libraries
• Comprehensive Testing: 87 unit tests with full rotation coverage

Performance Metrics

• Processing speed: Up to 6,259 pixels/ms
• OpenMP parallel loops: 25+
• Test coverage: 87/87 tests passing
• Rotation accuracy: 100% (all angles 0°-180°)

Building

Requirements

• CMake 3.10+
• C++23 compiler (GCC 11+, Clang 14+)
• OpenMP (optional but recommended)
• ImageMagick (optional, for PNG conversion)

Build Commands

./b  # Build project and run tests
./r  # Build and run main application
./v  # Build and run visual tests

Or manually:

mkdir build && cd build
cmake .. && make

Usage

Main Application

./Output/CppRectangleRecognition

Input: PGM grayscale images
Output: PNG images with detected shapes outlined

Test Suite

./Output/tests                # Run all unit tests
./Output/TestPerformance      # Run performance benchmarks
./Output/VisualTest          # Generate visual test images

Project Structure

CppRectangleRecognition/
├── Include/              # Header files
│   └── ShapeDetector/    # Detection algorithms
├── Source/               # Implementation files
├── Test/                 # Unit tests
├── Output/               # Executables and output images
├── build/                # Build artifacts
└── CMakeLists.txt        # Build configuration

Algorithm Overview

Detection Pipeline

1. Multi-Strategy Preprocessing

   • Standard contour detection
   • Enhanced edge detection
   • Morphological processing
   • Multi-threshold analysis
   • Edge-preserving filtering

2. Shape Analysis

   • Rectangle: Corner validation, parallel sides, rotation invariance
   • Sphere/Obloid: Circularity analysis, ellipticity metrics, radial consistency

3. Validation

   • Multi-level validation (strict/moderate/relaxed)
   • Moment-based shape classification
   • Geometric property verification

Configuration

Rectangle Detector

detector.SetMinArea(100.0);      // Minimum area threshold
detector.SetMaxArea(50000.0);    // Maximum area threshold
detector.SetApproxEpsilon(0.02); // Contour approximation precision

Sphere Detector

detector.SetMinRadius(10);       // Minimum radius
detector.SetMaxRadius(200);      // Maximum radius
detector.SetCircularityThreshold(0.7); // Circularity threshold

Performance Optimizations

• Compiler: -O3, -march=native, -flto, -ffast-math
• Parallelization: OpenMP on all critical loops
• Algorithm: Early exit conditions, pre-allocated caches
• Memory: Cache-friendly access patterns

Testing

The project includes comprehensive test coverage:

• Geometry primitives testing
• Image processing validation
• Rectangle detection (all rotations)
• Sphere/obloid detection
• Performance benchmarks
• Visual verification suite

Output Examples

Visual test suite generates test scenarios:

• visual_test_circles_only.png: Circle detection
• visual_test_rectangles_only.png: Rectangle detection
• visual_test_rotated_rectangles.png: Rotation testing
• visual_test_mixed_shapes.png: Multi-shape discrimination
• visual_test_complex_scene.png: Complex scene analysis

License

MIT License