Back to Aurora Vision Library website

You are here: Start » Function Reference » Computer Vision » 1D Edge Detection » ScanExactlyNEdges_Direct


Header: AVL.h
Namespace: avl
Module: MetrologyBasic

Locates a specified number of the strongest transitions between dark and bright pixels along a given path (without a scan map).

Applications: Very fast object detection (or presence verification) when the expected number of edges is clearly defined.


void avl::ScanExactlyNEdges_Direct
	const avl::Image& inImage,
	const avl::Path& inScanPath,
	atl::Optional<const avl::CoordinateSystem2D&> inScanPathAlignment,
	int inScanWidth,
	const avl::SamplingParams& inSamplingParams,
	const avl::EdgeScanParams& inEdgeScanParams,
	int inEdgeCount,
	avl::Selection::Type inEdgeSelection,
	float inMinDistance,
	atl::Optional<float> inMaxDistance,
	atl::Optional<const avl::LocalBlindness&> inLocalBlindness,
	atl::Conditional<atl::Array<avl::Edge1D> >& outEdges,
	atl::Conditional<atl::Array<avl::Gap1D> >& outGaps,
	atl::Optional<avl::Path&> outAlignedScanPath = atl::NIL,
	atl::Optional<avl::Profile&> outBrightnessProfile = atl::NIL,
	atl::Optional<avl::Profile&> outResponseProfile = atl::NIL,
	atl::Array<avl::Path>& diagSamplingPoints,
	float& diagSamplingStep


Name Type Range Default Description
Input value inImage const Image& Input image
Input value inScanPath const Path& Path along which the scan is performed
Input value inScanPathAlignment Optional<const CoordinateSystem2D&> NIL Adjusts the scan path to the position of the inspected object
Input value inScanWidth int 1 - 5 Width of the scan field in pixels
Input value inSamplingParams const SamplingParams& SamplingParams ( Interpolation: Bilinear SamplingStep: 1.0f SampleCount: Nil ) Parameters controlling the sampling process
Input value inEdgeScanParams const EdgeScanParams& EdgeScanParams ( ProfileInterpolation: Quadratic4 SmoothingStdDev: 0.6f MinMagnitude: 5.0f EdgeTransition: BrightToDark ) Parameters controlling the edge extraction process
Input value inEdgeCount int 0 - 1 Number of edges to be found
Input value inEdgeSelection Selection::Type Selection mode of the resulting edges
Input value inMinDistance float 0.0 - 0.0f Minimal distance between consecutive edges
Input value inMaxDistance Optional<float> 0.0 - NIL Maximal distance between consecutive edges
Input value inLocalBlindness Optional<const LocalBlindness&> NIL Defines conditions in which weaker edges can be detected in the vicinity of stronger edges
Output value outEdges Conditional<Array<Edge1D> >& Found edges
Output value outGaps Conditional<Array<Gap1D> >& Gaps between consecutive edges
Output value outAlignedScanPath Optional<Path&> NIL Transformed input path
Output value outBrightnessProfile Optional<Profile&> NIL Extracted image profile
Output value outResponseProfile Optional<Profile&> NIL Profile of the edge (derivative) operator response
Diagnostic input diagSamplingPoints Array<Path>& Array of paths each one containing the sampling points that contributes to a single value of the extracted profile
Diagnostic input diagSamplingStep float& Used distance between consecutive sampling points on the scan path

Optional Outputs

The computation of following outputs can be switched off by passing value atl::NIL to these parameters: outAlignedScanPath, outBrightnessProfile, outResponseProfile.

Read more about Optional Outputs.


The operation scans the image along inScanPath and finds a set of inEdgeCount image edges perpendicular to the path. If no subset (of inEdgeCount elements) of detected edges meets the requirements of inEdgeScanParams.minMagnitude, inMinDistance, inEdgeScanParams.edgeTransition then the outputs are set to NIL.

The optional parameter inScanPathAlignment defines a transform to be performed on the inScanPath so that the actual scan path (outAlignedScanPath) is adjusted to the position of the object, typically detected by one of Template Matching filters.

Note that in case of a scan path which is closed, the parameters controlling the distances between consecutive found objects do not control the distance between the first and the last of the found objects (counting from the beginning of the scan path).


  • Set inEdgeCount to the number of edges that are to be found (the N number).
  • Define inEdgeScanParams.EdgeTransition to detect a particular edge type, and only that type.
  • If the expected number of edges cannot be found, try decreasing inEdgeScanParams.MinMagnitude. Verify this with the values on the outResponseProfile output.
  • If consecutive edges are closer than 6 pixels apart, change inEdgeScanParams.ProfileInterpolation to Quadratic3.
  • Adjust inMinDistance (in pixels) to filter out false edges that appear in proximity to other edges.


ScanExactlyNEdges_Direct locates the edges across inScanPath (inEdgeCount = 61).


Read more about Local Coordinate Systems in Machine Vision Guide: Local Coordinate Systems.

This filter is a part of the 1D Edge Detection toolset. For a comprehensive introduction to this technique please refer to 1D Edge Detection and 1D Edge Detection - Subpixel Precision chapters of our Machine Vision Guide.

See Also

  • CreateScanMap – Precomputes a data object that is required for fast 1D edge detection.
  • ScanSingleEdge_Direct – Locates the strongest transition between dark and bright pixels along a given path (without a scan map).
  • ScanMultipleEdges_Direct – Locates multiple transitions between dark and bright pixels along a given path (without a scan map).
  • ScanExactlyNEdges – Locates a specified number of the strongest transitions between dark and bright pixels along a given path.