How astaranowicz/dcct works
Project Type and Purpose
• It provides an interactive command-line interface (CLI) in MATLAB to process RGB images and corresponding depth maps, fit geometric primitives (circles, ellipses, spheres), and compute calibration parameters via linear and non-linear minimization.
a. Host Environment
• MATLAB runtime / interpreter (scripts .m files execute here)
b. Entry Point & Configuration
• DCCT_v1_1.m – Main orchestration script (CLI menu, driver)
• DCCT_variables_SetupScript.m – Global default parameters (do not modify)
c. Data I/O
• Data folder – Raw inputs: RGBImage_ball_#.jpg and ball_#.txt (depth vectors)
• Save/Load routines – MATLAB .mat files for intermediate session data
d. RGB Image Processing Module
• f_RGB_imageProcessing.m – Interactive: rectangle selection → Canny edge → Hough circle → RANSAC ellipse fit → user validation
• Hough Transform for Circles toolbox (houghcircles.m and related)
• Ellipse-fitting functions (f_EllipseDirectFit.m, f_ellipseFit_points2Ellipse.m, etc.)
e. Depth Map Processing Module
• f_Depth_imageProcessing.m – Interactive: rectangle selection → depth thresholding → pixel filtering → user validation
• Depth-to-XYZ conversions (f_depth2XYZ.m, f_DepthImage2Depthmap.m)
f. Calibration Core Module
• Linear estimation – “6-point” and “sphere” algorithms (f_sixPointAlgo.m, f_minSphere.m, f_sphereFit_points2Sphere.m, etc.)
• Non-linear minimization – f_nonLinearMinization_6pnt_Conic.m (refines intrinsic, extrinsic, distortion)
• Conic toolbox (conic/*.m) – algebraic routines for conic sections used in sphere/ellipse estimation
• Epipolar Geometry Toolbox (EGT_v2p0) – external dependency for epipolar computations
g. Output & Visualization
• Figures of fitted ellipses on images, depth overlays
• Final calibration parameters (intrinsics, extrinsics, distortion coefficients) written to workspace or saved to file
Component Interactions & Data Flow
1) User launches DCCT_v1_1.m in MATLAB
2) Main script reads DCCT_variables_SetupScript for default settings
3) User chooses “new session” or “load previous data”
4) For each RGB-D frame:
– Call RGB Image Processing → returns 2D ellipse parameters
– Call Depth Map Processing → returns 3D point cloud of sphere surface
– Store results in internal data structure
5) Once all frames are processed, pass collected 2D–3D correspondences to Calibration Core
6) Calibration Core performs linear estimation, then non-linear minimization
7) Generate output: calibration parameters, plots
8) Optionally save session data in Data folder
Architectural Patterns & Design Principles
• Modular pipeline: clear separation between data acquisition (processing) and algorithmic calibration
• Interactive feedback loop: user-in-the-loop validation for robust feature extraction
• Extensibility points: external toolboxes (conic, EGT, Hough) plugged in as libraries
• Single responsibility: each function/script focuses on one stage
Detailed Element Checklist
• Host: MATLAB Environment
• Main Script: DCCT_v1_1.m
• Config: DCCT_variables_SetupScript.m
• RGB Processor: f_RGB_imageProcessing + Hough/Ellipse libs
• Depth Processor: f_Depth_imageProcessing + depth conversion
• Calibration Engine: linear estimators, f_nonLinearMinization_6pnt_Conic
• External Toolboxes: conic/, EGT_v2p0/, HoughTransform
• Data Store: Data/ folder + save/load routines
• Outputs: Calibration parameters + Visualizations