Struct DetectorConfig 

pub struct DetectorConfig {

    pub threshold_tile_size: usize,
    pub threshold_min_range: u8,
    pub enable_sharpening: bool,
    pub enable_adaptive_window: bool,
    pub threshold_min_radius: usize,
    pub threshold_max_radius: usize,
    pub adaptive_threshold_constant: i16,
    pub adaptive_threshold_gradient_threshold: u8,
    pub quad_min_area: u32,
    pub quad_max_aspect_ratio: f32,
    pub quad_min_fill_ratio: f32,
    pub quad_max_fill_ratio: f32,
    pub quad_min_edge_length: f64,
    pub quad_min_edge_score: f64,
    pub subpixel_refinement_sigma: f64,
    pub segmentation_margin: i16,
    pub segmentation_connectivity: SegmentationConnectivity,
    pub upscale_factor: usize,
    pub decimation: usize,
    pub nthreads: usize,
    pub decoder_min_contrast: f64,
    pub refinement_mode: CornerRefinementMode,
    pub decode_mode: DecodeMode,
    pub max_hamming_error: u32,
    pub huber_delta_px: f64,
    pub tikhonov_alpha_max: f64,
    pub sigma_n_sq: f64,
    pub structure_tensor_radius: u8,
    pub gwlf_transversal_alpha: f64,
    pub quad_max_elongation: f64,
    pub quad_min_density: f64,
    pub quad_extraction_mode: QuadExtractionMode,
}

Pipeline-level configuration for the detector.

These settings affect the fundamental behavior of the detection pipeline and are immutable after the Detector is constructed. Use the builder pattern for ergonomic construction.

Fields

threshold_tile_size: usize

Tile size for adaptive thresholding (default: 4). Larger tiles are faster but less adaptive to local contrast.

threshold_min_range: u8

Minimum intensity range in a tile to be considered valid (default: 10). Tiles with lower range are treated as uniform (no edges).

enable_sharpening: bool

Enable Laplacian sharpening to enhance edges for small tags (default: true).

enable_adaptive_window: bool

Enable adaptive threshold window sizing based on gradient (default: true).

threshold_min_radius: usize

Minimum threshold window radius for high-gradient regions (default: 2 = 5x5).

threshold_max_radius: usize

Maximum threshold window radius for low-gradient regions (default: 7 = 15x15).

adaptive_threshold_constant: i16

Constant subtracted from local mean in adaptive thresholding (default: 3).

adaptive_threshold_gradient_threshold: u8

Gradient magnitude threshold above which the minimum window radius is used (default: 40).

quad_min_area: u32

Minimum quad area in pixels (default: 16).

quad_max_aspect_ratio: f32

Maximum aspect ratio of bounding box (default: 3.0).

quad_min_fill_ratio: f32

Minimum fill ratio (pixel count / bbox area) (default: 0.3).

quad_max_fill_ratio: f32

Maximum fill ratio (default: 0.95).

quad_min_edge_length: f64

Minimum edge length in pixels (default: 4.0).

quad_min_edge_score: f64

Minimum edge alignment score (0.0 to 1.0)

subpixel_refinement_sigma: f64

PSF blur factor for subpixel refinement (e.g., 0.6)

segmentation_margin: i16

Minimum deviation from threshold for a pixel to be connected in threshold-model CCL (default: 2).

segmentation_connectivity: SegmentationConnectivity

Segmentation connectivity (4-way or 8-way).

upscale_factor: usize

Factor to upscale the image before detection (1 = no upscaling). Increasing this to 2 allows detecting smaller tags (e.g., < 15px) at the cost of processing speed (O(N^2)). Nearest-neighbor interpolation is used.

decimation: usize

Decimation factor for preprocessing (1 = no decimation).

nthreads: usize

Number of threads for parallel processing (0 = auto).

decoder_min_contrast: f64

Minimum contrast range for Otsu-based bit classification (default: 20.0). For checkerboard patterns with densely packed tags, lower values (e.g., 10.0) can improve recall on small/blurry tags.

refinement_mode: CornerRefinementMode

Strategy for refining corner positions (default: Edge).

decode_mode: DecodeMode

Decoding mode (Hard vs Soft).

max_hamming_error: u32

Maximum number of Hamming errors allowed for tag decoding (default: 2). Higher values increase recall but also increase false positive rate in noise.

huber_delta_px: f64

Huber delta for LM reprojection (pixels) in Fast mode. Residuals beyond this threshold are down-weighted linearly. 1.5 px is a standard robust threshold for sub-pixel corner detectors.

tikhonov_alpha_max: f64

Maximum Tikhonov regularisation alpha (px^2) for ill-conditioned corners in Accurate mode. Controls the gain-scheduled regularisation of the Structure Tensor information matrix on foreshortened tags.

sigma_n_sq: f64

Pixel noise variance (sigma_n^2) assumed for the Structure Tensor covariance model in Accurate mode. Typical webcams: ~4.0.

structure_tensor_radius: u8

Radius (in pixels) of the window used for Structure Tensor computation in Accurate mode. A radius of 2 yields a 5x5 window. Smaller values (1) are better for small tags; larger (3-4) for noisy images. Validation caps this at 8 to keep the covariance kernel stack-only.

gwlf_transversal_alpha: f64

Alpha parameter for GWLF adaptive transversal windowing. The search band is set to +/- max(2, alpha * edge_length).

quad_max_elongation: f64

Maximum elongation (λ_max / λ_min) allowed for a component before contour tracing. 0.0 = disabled. Recommended: 15.0 to reject thin lines and non-square blobs.

quad_min_density: f64

Minimum pixel density (pixel_count / bbox_area) required to pass the moments gate. 0.0 = disabled. Recommended: 0.2 to reject sparse/noisy regions.

quad_extraction_mode: QuadExtractionMode

Quad extraction mode: legacy contour tracing (default) or EDLines.

Implementations

impl DetectorConfig

pub fn builder() -> DetectorConfigBuilder

Create a new builder for DetectorConfig.

pub fn validate(&self) -> Result<(), ConfigError>

Validate the configuration, returning an error if any parameter is out of range.

Errors

Returns [ConfigError] if any parameter violates its constraints.

impl DetectorConfig

pub fn from_profile_json(json: &str) -> Result<Self, ConfigError>

Load a user-supplied profile from a JSON string.

Returns [ConfigError::ProfileParse] for malformed JSON or unknown fields (the serde deserializer rejects unknown keys), and any validation error from DetectorConfig::validate for configurations that fail cross-group compatibility checks (e.g. EdLines + Erf).

Errors

See above: parse failure and post-parse validation failure.

pub fn from_profile(name: &str) -> Self

Load one of the three shipped profiles by name.

Accepts "standard", "grid", or "high_accuracy".

Panics

Panics on an unknown profile name — this is a programming error against a closed set of three compile-time-embedded profiles. Panics on a malformed embedded JSON, which would be a build error caught by the profile_loading integration test.

Trait Implementations

impl Clone for DetectorConfig

fn clone(&self) -> DetectorConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for DetectorConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for DetectorConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for DetectorConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for DetectorConfig

fn eq(&self, other: &DetectorConfig) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for DetectorConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Copy for DetectorConfig

impl StructuralPartialEq for DetectorConfig