Operator Reference

learn_ndim_normT_learn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm (Operator)

learn_ndim_normT_learn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm — Construct classes for class_ndim_normclass_ndim_normClassNdimNormClassNdimNormclass_ndim_norm.

Signature

learn_ndim_norm(Foreground, Background, Image : : Metric, Distance, MinNumberPercent : Radius, Center, Quality)

Description

learn_ndim_normlearn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm generates classification clusters from the region ForegroundForegroundForegroundforegroundforeground and the corresponding gray values in the multi-channel image ImageImageImageimageimage, which can be used in class_ndim_normclass_ndim_normClassNdimNormClassNdimNormclass_ndim_norm. BackgroundBackgroundBackgroundbackgroundbackground determines a class of pixels not to be found in class_ndim_normclass_ndim_normClassNdimNormClassNdimNormclass_ndim_norm. This parameter may be empty (empty object).

The parameter DistanceDistanceDistancedistancedistance determines the maximum distance RadiusRadiusRadiusradiusradius of the clusters. It describes the minimum distance between two cluster centers. If the parameter DistanceDistanceDistancedistancedistance is small the (small) hyper-cubes or hyper-spheres can approximate the feature space well. Simultaneously the runtime during classification increases.

The ratio of the number of pixels in a cluster to the total number of pixels (in percent) must be larger than the value of MinNumberPercentMinNumberPercentMinNumberPercentminNumberPercentmin_number_percent, otherwise the cluster is not returned. MinNumberPercentMinNumberPercentMinNumberPercentminNumberPercentmin_number_percent serves to eliminate outliers in the training set. If it is chosen too large many clusters are suppressed.

Two different clustering procedures can be selected: The minimum Euclidean distance algorithm (n-dimensional hyper-spheres) and the maximum algorithm (n-dimensional hyper-cubes) for describing the pixels of the image to classify in the n-dimensional histogram (parameter MetricMetricMetricmetricmetric). The Euclidean metric usually yields the better results, but takes longer to compute. The parameter QualityQualityQualityqualityquality returns the quality of the clustering. It is a measure of overlap between the rejection class and the classificator classes. Values larger than 0 denote the corresponding ratio of overlap. If no rejection region is given, its value is set to 1. The regions in BackgroundBackgroundBackgroundbackgroundbackground do not influence on the clustering. They are merely used to check the results that can be expected.

From a user's point of view the key difference between learn_ndim_normlearn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm and learn_ndim_boxlearn_ndim_boxLearnNdimBoxLearnNdimBoxlearn_ndim_box is that in the latter case the rejection class affects the classification process itself. Here, a hyper plane is generated that separates ForegroundForegroundForegroundforegroundforeground and BackgroundBackgroundBackgroundbackgroundbackground classes, so that no points in feature space are classified incorrectly. As for learn_ndim_normlearn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm, however, an overlap between ForegroundForegroundForegroundforegroundforeground and BackgroundBackgroundBackgroundbackgroundbackground class is allowed. This has its effect on the return value QualityQualityQualityqualityquality. The larger the overlap, the smaller this value.

Execution Information

Multithreading type: exclusive (runs in parallel only with independent operators).
Multithreading scope: global (may be called from any thread).
Processed without parallelization.

Parameters

ForegroundForegroundForegroundforegroundforeground (input_object) region(-array) → object

Foreground pixels to be trained.

BackgroundBackgroundBackgroundbackgroundbackground (input_object) region(-array) → object

Background pixels to be trained (rejection class).

ImageImageImageimageimage (input_object) (multichannel-)image(-array) → object (byte)

Multi-channel training image.

MetricMetricMetricmetricmetric (input_control) string → (string)

Metric to be used.

Default: 'euclid' "euclid" "euclid" "euclid" "euclid"

List of values: 'euclid'"euclid""euclid""euclid""euclid", 'maximum'"maximum""maximum""maximum""maximum"

DistanceDistanceDistancedistancedistance (input_control) number → (real / integer)

Maximum cluster radius.

Default: 10.0

Suggested values: 1.0, 2.0, 3.0, 4.0, 6.0, 8.0, 10.0, 13.0, 17.0, 24.0, 30.0, 40.0

Value range: 1.0 ≤ Distance Distance Distance distance distance (lin)

Minimum increment: 0.01

Recommended increment: 1.0

MinNumberPercentMinNumberPercentMinNumberPercentminNumberPercentmin_number_percent (input_control) number → (real / integer)

The ratio of the number of pixels in a cluster to the total number of pixels (in percent) must be larger than MinNumberPercent (otherwise the cluster is not output).

Default: 0.01

Suggested values: 0.001, 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, 5.0, 10.0

Value range: 0.0 ≤ MinNumberPercent MinNumberPercent MinNumberPercent minNumberPercent min_number_percent ≤ 100.0 (lin)

Minimum increment: 0.01

Recommended increment: 0.1

RadiusRadiusRadiusradiusradius (output_control) real-array → (real)

Cluster radii or half edge lengths.

CenterCenterCentercentercenter (output_control) real-array → (real)

Coordinates of all cluster centers.

QualityQualityQualityqualityquality (output_control) real → (real)

Overlap of the rejection class with the classified objects (1: no overlap).

Assertion: 0 <= Quality && Quality <= 1

Result

learn_ndim_normlearn_ndim_normLearnNdimNormLearnNdimNormlearn_ndim_norm returns 2 ( H_MSG_TRUE) if all parameters are correct. The behavior with respect to the input images can be determined by setting the values of the flags 'no_object_result'"no_object_result""no_object_result""no_object_result""no_object_result" and 'empty_region_result'"empty_region_result""empty_region_result""empty_region_result""empty_region_result" with set_systemset_systemSetSystemSetSystemset_system. If necessary, an exception is raised.

Possible Predecessors

min_max_graymin_max_grayMinMaxGrayMinMaxGraymin_max_gray, sobel_ampsobel_ampSobelAmpSobelAmpsobel_amp, binomial_filterbinomial_filterBinomialFilterBinomialFilterbinomial_filter, gauss_filtergauss_filterGaussFilterGaussFiltergauss_filter, reduce_domainreduce_domainReduceDomainReduceDomainreduce_domain, diff_of_gaussdiff_of_gaussDiffOfGaussDiffOfGaussdiff_of_gauss

Possible Successors

class_ndim_normclass_ndim_normClassNdimNormClassNdimNormclass_ndim_norm, connectionconnectionConnectionConnectionconnection, dilation1dilation1Dilation1Dilation1dilation1, erosion1erosion1Erosion1Erosion1erosion1, openingopeningOpeningOpeningopening, closingclosingClosingClosingclosing, rank_regionrank_regionRankRegionRankRegionrank_region, shape_transshape_transShapeTransShapeTransshape_trans, skeletonskeletonSkeletonSkeletonskeleton

Alternatives

learn_ndim_boxlearn_ndim_boxLearnNdimBoxLearnNdimBoxlearn_ndim_box, learn_class_boxlearn_class_boxLearnClassBoxLearnClassBoxlearn_class_box

References

P. Haberäcker, “Digitale Bildverarbeitung”; Hanser-Studienbücher, München, Wien, 1987

Module

Foundation

Operators