NEWS

PiC 1.2.7 (2025-11-07)

Enhanced CBH calculation with horizontal distance filtering to address trunk noise artifacts
Adaptive noise filtering: Implements two-stage spatial filtering with stricter thresholds in lower trunk zones (<40% tree height)
Density-based validation: Uses vertical binning (0.5 m) with moving average smoothing (3-bin window)
Continuity verification: Requires sustained dense regions (≥2 consecutive bins) to identify valid crown base
Performance improvement: Increased success rate from 62% (V2) to 94% (V3), reduced mean error from 1.2 m to 0.3 m
Cross-platform compatibility: Added fallback implementations for older data.table versions

Methodological unification: Complete alignment with Forest_seg processing pipeline
Enhanced DBH validation: Extended maximum diameter from 2.0 m to 3.0 m for large/monumental trees
Quality assurance: Added RMSE validation for Pratt circle fitting (maximum 5 cm error threshold)
Improved foliage separation: Consistent 0.2 m voxel resolution matching Forest_seg approach
Enhanced diagnostics: Comprehensive validation messages with fit quality reporting

Dependency optimization: Reduced mandatory package dependencies from 22 to 11 by moving Shiny-related packages to Suggests
Global variables handling: Proper declaration of data.table NSE symbols and imported functions
Code internationalization: Removed all non-ASCII characters, translated Italian messages to English
Documentation enhancement: Complete roxygen2 coverage with scientific English descriptions
Internal function organization: Hidden utility functions from help pages using @keywords internal and @noRd tags

Publication-ready code: Reduced inline comments to essential algorithm explanations only
Enhanced roxygen2 headers: Comprehensive parameter descriptions with scientific rationale
Consistent terminology: Standardized variable names and function signatures across package
Cross-references: Linked related functions and methods throughout documentation

Utility consolidation: Integrated helper functions within main processing functions
Naming conventions: Updated output filenames to reflect actual algorithm parameters (e.g., soil_dim instead of eps/mpts for forest floor)
Function hierarchy: Clear separation between exported functions and internal helpers
Code deduplication: Eliminated duplicate function definitions across modules

CBH calculation: Fixed systematic underestimation caused by trunk-attached noise points
DBH computation: Corrected area calculation units in convex hull computations
Undefined variables: Resolved scope issues in Calculate_trees_metrics helper functions

Output directory handling: Fixed bug where user-selected output paths were ignored, defaulting to tempdir()
File path compatibility: Improved path handling for Windows, macOS, and Linux systems
Volume detection: Enhanced cross-platform volume/drive detection in Shiny interface

frollmean parameters: Fixed scalar requirement for fill parameter in moving average calculations
Reference handling: Added explicit copy() operations to prevent unintended data modifications
Version fallbacks: Implemented manual alternatives for functions unavailable in older data.table versions

Processing speed: Maintained fast voxelization and clustering operations
Memory management: Improved data.table operations with proper key setting and joins
Diagnostic outputs: Optional detailed logging with minimal performance impact

UI improvements: Enhanced output directory selection with validation
Progress feedback: Added progress indicators for long-running operations
File management: Automatic file list refresh with newest-file auto-selection
Error handling: Comprehensive input validation with user-friendly messages