**
preprocess_dl_classifier_images (Procedure)
**
简短描述
preprocess_dl_classifier_images — Preprocess images for deep-learning-based classification training and inference.
*预处理图像,为基于深度学习的分类训练和推论做准备
描述
This procedure preprocesses the provided images given by Images so that they can be handled by train_dl_classifier_batch and apply_dl_classifier_batch.
*本操作对由Image变量提供的图像进行预处理,使得它们可以被train_dl_classifier_batch和apply_dl_classifier_batch操作
Note that depending on the data set, additional preprocessing steps might be beneficial.
*需要注意的是,根据数据集的不同,额外的预处理步骤可能是有益的
The preprocessed images are returned in ImagePreprocessed. The network’s image requirements are provided by the DLClassifierHandle.
*经过预处理的图像由ImagePreprocessed返回。网络对于图像的要求由DLClassifierHandle给出
The deep-learning-based classifier has certain requirements on the images. In this procedure, the width, height and gray value range of each image is adapted accordingly.
*基于深度学习的分类器对于图像有确切要求。在本操作中,图像的宽度,高度和灰度值范围据此相适应
It is required that all images passed to train_dl_classifier_batch and apply_dl_classifier_batch are provided as returned by this procedure.
*所有要进行train_dl_classifier_batch 和 apply_dl_classifier_batch操作的图像都必须是本操作的返回值
For the training, for example, you can do this step at the beginning of your program for all images, or alternatively for each batch that is trained separately.
*对于训练而言,比如说,你可以在程序开头对所有图像都作本操作,或者对于每一批分别训练的图像分别做本操作
参数
Images (input_object) (multichannel-)image(-array) → object (byte / real)
Images that are preprocessed.
ImagesPreprocessed (output_object) multichannel-image(-array) → object (real)
Preprocessed images.
GenParamName (input_control) attribute.name(-array) → (string)
Name of the generic parameter.
Default Value: []
建议值: ‘contrast_normalization’, ‘domain_handling’
GenParamValue (input_control) attribute.value(-array) → (integer / real / string)
Value of the generic parameter.
Default Value: []
建议值: ‘true’, ‘false’, ‘full_domain’, ‘crop_domain’
DLClassifierHandle (input_control) dl_classifier → (integer)
Handle of the deep-learning-based classifier.