{"name": "ImageQualityToolset", "package_name": "image_quality_toolset", "version": "1.2", "experimental": false, "qgis_min": "3.38.0", "qgis_max": "4.99.0", "downloads": 26, "uploaded_by": "rdegoul", "upload_datetime": "2026-11-03T09:18:50.678266", "changelog": "ImageQualityToolset 1.2\n- Fixed the Bandit security findings that led to the withdrawal of 1.1: the dependency bundle is now downloaded over https only (plain http is accepted on loopback, for the tests), and failed version lookups are logged instead of being silently ignored.\n- Bandit security audit added to the pre-commit hooks and to the CI.\n- Clearer error messages instead of tracebacks on an unusable ROI or input, and on a figure that cannot be saved.\n- Otherwise identical to 1.1.\nImageQualityToolset 1.1\n- QGIS 4 / Qt 6 support (also QGIS 3.38+).\n- Dependencies (loess, scikit-image) now come from a prebuilt bundle downloaded at first run and checked against a checksum, instead of pip. Status and errors are shown in the results panel; offline install is possible by dropping the bundle in the plugin folder.\n- Fixed the Bridge method on recent NumPy versions, and the MTF methods crashing on macOS.\n- Fixed the Knife Edge and Bridge methods failing with SCALE=1 (raw DN, the default): the ESF outlier cleaning now uses a threshold relative to the ESF range, so results no longer depend on the scale.\n- Fixed the Bridge method failing when the ROI polygon does not fill its bounding box (NaN pixels), and an input rotation angle of 0 being ignored.\n- A ROI with no CRS is accepted (its coordinates are taken in the raster CRS): a raw, non-georeferenced image and its ROI overlap in QGIS without choosing a CRS.\n- Fixed the results panel error \"wrapped C/C++ object of type QgsVectorLayer has been deleted\" when the ROI is given as a file.\nImageQualityToolset 1.0 - Initial release.\n- MTF Knife Edge Method: computes MTF from a sharp-edge ROI with parametric (sigmoid, tanh, Fermi, erf, Gaussian-exponential) and non-parametric (Loess, polynomial) ESF models. Reports MTF@Nyquist, MTF30, MTF50, FWHM, RER, HEE, SNR, and R\u00b2.\n- MTF Bridge (Pulse) Method: computes MTF from a pulse/bridge target ROI.\n- SNR Variogram Method: estimates SNR combining a Histogram method and a Variogram method with per-pixel local SNR values.\n- Outputs: synthetic diagnostic PNG figure, detailed analysis PNG figure, and cumulative GeoPackage report (one polygon feature per run).\n- QGIS Processing Toolbox integration and standalone command-line interface via INI configuration files.\n- Dual-license model: GPL v3 for the QGIS plugin layer, Apache 2.0 for the core signal-processing algorithms.", "external_deps": "", "download_url": "https://plugins.qgis.org/plugins/image_quality_toolset/version/1.2/download/"}