train_variation_modelT_train_variation_modelTrainVariationModelTrainVariationModel (Operator)

Name

train_variation_modelT_train_variation_modelTrainVariationModelTrainVariationModel — Train a variation model.

Signature

train_variation_model(Images : : ModelID : )

train_variation_modeltrain_variation_modelTrainVariationModelTrainVariationModelTrainVariationModel trains the variation model that is passed in ModelIDModelIDModelIDModelIDmodelID with one or more images, which are passed in ImagesImagesImagesImagesimages.

As described for create_variation_modelcreate_variation_modelCreateVariationModelCreateVariationModelCreateVariationModel, a variation model that has been created using the mode 'standard'"standard""standard""standard""standard" can be trained iteratively, i.e., as soon as images of good objects become available, they can be trained with train_variation_modeltrain_variation_modelTrainVariationModelTrainVariationModelTrainVariationModel. The ideal image of the object is computed as the mean of all previous training images and the images that are passed in ImagesImagesImagesImagesimages. The corresponding variation image is computed as the standard deviation of the training images and the images that are passed in ImagesImagesImagesImagesimages.

If the variation model has been created using the mode 'robust'"robust""robust""robust""robust", the model cannot be trained iteratively, i.e., all training images must be accumulated using concat_objconcat_objConcatObjConcatObjConcatObj and be trained with train_variation_modeltrain_variation_modelTrainVariationModelTrainVariationModelTrainVariationModel in a single call. If any images have been trained previously, the training information of the previous call is discarded. The image of the ideal object is computed as the median of all training images passed in ImagesImagesImagesImagesimages. The corresponding variation image is computed as a suitably scaled median absolute deviation of the training images and the median image.

Attention

At most 65535 training images can be trained.

Execution Information

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

This operator modifies the state of the following input parameter:

ModelIDModelIDModelIDModelIDmodelID

The value of this parameter may not be shared across multiple threads without external synchronization.

Parameters

ImagesImagesImagesImagesimages (input_object) singlechannelimage(-array) → object (byte / int2 / uint2)

Images of the object to be trained.

ModelIDModelIDModelIDModelIDmodelID (input_control, state is modified) variation_model → (handle)

ID of the variation model.

Example (HDevelop)

open_framegrabber ('File', 1, 1, 0, 0, 0, 0, 'default', -1, \
                   'default', -1, 'default', 'model.seq', 'default', \
                   -1, -1, AcqHandle)
grab_image (Image, AcqHandle)
get_image_pointer1 (Image, Pointer, Type, Width, Height)
dev_display (Image)
draw_region (Region, WindowHandle)
reduce_domain (Image, Region, ImageReduced)
area_center (Region, Area, RowRef, ColumnRef)
create_shape_model (ImageReduced, 4, 0, rad(360), rad(1), 'none', \
                    'use_polarity', 40, 10, TemplateID)
create_variation_model (Width, Height, Type, 'standard', ModelID)
for K := 1 to 100 by 1
    grab_image (Image, AcqHandle)
    find_shape_model (Image, TemplateID, 0, rad(360), 0.5, 1, 0.5, \
                      'true', 4, 0.9, Row, Column, Angle, Score)
    if (|Score| == 1)
        vector_angle_to_rigid (Row, Column, Angle, RowRef, \
                               ColumnRef, 0, HomMat2D)
        affine_trans_image (Image, ImageTrans, HomMat2D, 'constant', \
                            'false')
        train_variation_model (ImageTrans, ModelID)
    endif
endfor
prepare_variation_model (ModelID, 10, 4)
write_region (Region, 'model.reg')
write_shape_model (TemplateID, 'model.shm')
write_variation_model (ModelID, 'model.var')
close_framegrabber (AcqHandle)

Result

train_variation_modeltrain_variation_modelTrainVariationModelTrainVariationModelTrainVariationModel returns 2 (H_MSG_TRUE) if all parameters are correct.