Linear Live View App

# line_camera: camera-based measurement of plotted function parameters **Open-source software and supporting data** for recovering plotted function class and parameters from graph paper imaged by a commodity webcam. After four-point homographic rectification, the software traces ink strokes and compares linear, quadratic, and sinusoidal models using penalized RMSE, with an arc-length quality gate that flags grid-ink contamination. This deposit packages the Python application, ground-truth plots, validation logs, baseline comparisons, uncertainty study outputs, protocols, and figure assets needed to reproduce the reported results. **Authors:** Phillip M. Wu et al. **Institution:** Institute of Physics, Academia Sinica **Software:** `line_camera` (Python 3; OpenCV, NumPy) --- ## Package contents | Path | Description ||------|-------------|| `LICENSE` | MIT License (code, data, and documentation in this deposit) || `code/line_camera.py` | Live camera measurement app (homography, stroke tracing, shape-aware fitting) || `code/scripts/` | Batch analysis, baselines, uncertainty Monte Carlo, figure generation || `code/requirements.txt` | Python dependencies || `code/linear.m` | MATLAB script used to generate reference plot families || `data/static_plots/` | 15 labeled JPEG validation images (8 primary + 7 `*a` variants) || `data/ground_truth.json` | Reference class and equation parameters for each plot || `data/MST_TRIAL_MATRIX.json` | Planned live-camera trial schedule (filename retained for compatibility) || `data/WebPlotDigitizer_results/` | Digitized point CSVs from WebPlotDigitizer baseline || `data/matlab_figs/` | MATLAB `.fig` sources for primary plots || `results/camera_log.jsonl` | **295** live-camera trials (primary validation log) || `results/camera_captures/` | Warped frames linked from `camera_log.jsonl` || `results/validation_results.json` | Static JPEG batch analysis || `results/phase3_camera_summary.json` | Aggregated live-camera (Phase-3) metrics || `results/uncertainty_jitter.json` | Corner-jitter Monte Carlo uncertainty || `results/baseline_hough.json` | Naive Hough line baseline || `results/baseline_webplotdigitizer.*` | WebPlotDigitizer baseline summary || `results/MST_ANALYSIS_SUMMARY.md` | Aggregated validation summary (filename retained for compatibility) || `results/figures/` | Phase-3 and uncertainty analysis plots || `results/overlays/` | Static analysis overlay images || `figures/report_figures/` | Report figure PNGs || `figures/mst_submission/` | Additional packaged figure set (filename retained for compatibility) || `protocols/` | Camera, uncertainty, baseline, and validation protocols || `MANIFEST.txt` | Full file list for this deposit | > **Note on filenames:** Some paths still use an `MST_` prefix or folder name from an earlier packaging iteration. They describe the same validation artifacts; content is not tied to any particular journal. --- ## Key reported metrics (from `results/MST_ANALYSIS_SUMMARY.md`) - Static JPEG classification: **100%** (15/15)- Live-camera trials: **295**; overall classification **72.2%** (213/295)- QC-pass (stroke arc length < 100 axis units): **86.7%** (182/210)- Corner-jitter uncertainty: mean slope SD ≈ **0.020** at σ = 2 px- WebPlotDigitizer: **7/8** class correct; mean line slope error **0.34%** --- ## Quick start (reproduce static + baselines) ```bash# From a working copy of this deposit (or the project tree with the same layout)python3 -m venv .venv.venv/bin/pip install -r code/requirements.txt # Place static plots and ground_truth where scripts expect them, or run from the# original Linear/ project layout documented in protocols/README_camera.md .venv/bin/python code/scripts/analyze_validation_set.py.venv/bin/python code/scripts/baseline_hough_static.py.venv/bin/python code/scripts/uncertainty_corner_jitter.py --samples 200.venv/bin/python code/scripts/process_webplotdigitizer_baseline.py.venv/bin/python code/scripts/analyze_mst_validation.py``` Live camera logging (requires webcam and printed plots): ```bash.venv/bin/python code/line_camera.py --validate# After calibration, press l to log each trial``` See `protocols/CAMERA_VALIDATION.md` and `protocols/README_camera.md` for the full procedure. Additional detailed run sheets live under `protocols/MST_*.md` (historical filenames; same methods). **Note:** Analysis scripts assume the original project layout (`Linear/line_camera.py`, `Linear/validation/...`). When unpacking this Zenodo deposit alone, either restore that layout or adjust paths in the scripts. --- ## Suggested Zenodo / GitHub metadata - **Title:** line_camera: supporting data and code for camera-based measurement of plotted function parameters - **Upload type:** Software (or Dataset + Software) - **Description:** Open-source Python tool for live webcam measurement of plotted linear, quadratic, and sinusoidal functions on graph paper, with shape-aware model selection, validation logs, and uncertainty protocols. - **Keywords:** graph digitization; camera measurement; homography; uncertainty; computer vision; curve fitting; open-source software; scientific software - **License:** MIT (see `LICENSE`) - **Related identifiers:** 10.5281/zenodo.21386582 --- ## Citation > Wu, P. M. (2026). *line_camera: camera-based measurement of plotted function parameters* (software and supporting data). Institute of Physics, Academia Sinica. [10.5281/zenodo.21386582] --- ## Excluded from this deposit The following project files were **not** copied (not required for reproducibility of the numerical results): - Local Python virtualenv (`.venv`)- Manuscript drafts and cover letters- Large ad-hoc development screenshots- Unrelated reference PDFs- Opportunistic `captures/` frames outside the Phase-3 `camera_log.jsonl` set --- ## License This software and the accompanying data, protocols, and figures in this deposit are released under the **MIT License**. See `LICENSE` for the full text. Third-party tools used at runtime (e.g. OpenCV, NumPy, MATLAB for generating reference plots, WebPlotDigitizer for baseline CSVs) remain under their own licenses; only the original `line_camera` materials in this package are covered by the MIT grant above. --- ## Contact Phillip M. Wu — Institute of Physics, Academia Sinica, Taipei, Taiwan R.O.C.