Slice Defender

Technical Overview

System Architecture

• End-to-end real-time pipeline: Webcam → Vision Processing → Coordinate Mapping → 3D Rendering
• Tight coupling between OpenCV frame processing and Qt event loop via QTimer (~30 FPS)

Computer Vision Pipeline (OpenCV)

• Frame acquisition using VideoCapture
• Color-space conversion RGB → YCrCb
• Adaptive contrast enhancement using CLAHE on Cr/Cb channels
• Skin mask extraction via thresholding
• Noise reduction using morphological opening/closing
• Contour filtering to isolate hand region
• Distance Transform to estimate palm center
• Haar Cascade (detectMultiScale) as a fallback detection mechanism

Interaction & Mapping

• Palm center coordinates normalized and mirrored
• 2D camera space mapped to 3D world coordinates

Rendering & UI (Qt + OpenGL)

• Real-time 3D sword movement in OpenGL scene
• Qt-based GUI (camera preview, score, timer)
• Modular separation between vision, UI, and rendering layers
• Custom OpenGL scene embedded in Qt (QOpenGLWidget) with a manually designed 3D background
• Dynamic projectile system: multiple shapes spawned at random positions following projectile-motion trajectories
• Sword–projectile interaction model: sword position driven by palm tracking and mapped to 3D world space
• Collision detection based on cylindrical hit volume and radius-threshold intersection (||P_sword - P_proj|| < r)
• Successful intersections trigger explosion effects and score updates
• Qt-based HUD integrating camera preview, timer, score, with clean separation between vision, game logic, and rendering layers

Performance

• C++17 implementation
• Stable real-time interaction at ~30 FPS