Struct GroupCropAnalyzer 

pub struct GroupCropAnalyzer;

Group crop region analyzer using Tukey fence for outlier removal

Implementations

impl GroupCropAnalyzer

pub fn decide_group_crop_region( bounding_boxes: &[PageBoundingBox], ) -> GroupCropRegion

Decide the optimal crop region for a group of pages using Tukey fence

Algorithm:

1. Collect bounding boxes from all pages
2. Calculate Q1, Q3, IQR for each edge (left, top, right, bottom)
3. Apply Tukey fence (k=1.5) to identify outliers
4. Remove pages where ANY edge is an outlier
5. Calculate median of inliers for final crop region

pub fn unify_odd_even_regions( bounding_boxes: &[PageBoundingBox], ) -> UnifiedCropRegions

Unify crop regions for odd and even page groups

pub fn unify_and_expand_regions( bounding_boxes: &[PageBoundingBox], margin_percent: u32, max_width: u32, max_height: u32, ) -> UnifiedCropRegions

Unify crop regions with Y coordinate unification, margin expansion, and size limits

This function implements the full C# algorithm:

1. Calculate separate crop regions for odd/even pages
2. Unify Y coordinates (use min top, max bottom)
3. Expand width/height by margin_percent
4. Center the expansion
5. Clamp to image bounds

pub fn detect_text_bounding_box( image_path: &Path, background_threshold: u8, ) -> Result<ContentRect>

Detect text bounding box from image using edge detection

This function detects the content area by finding non-background pixels and returning the minimal bounding box.

pub fn detect_all_bounding_boxes( image_paths: &[PathBuf], background_threshold: u8, ) -> Vec<PageBoundingBox>

Detect bounding boxes for all pages in parallel