protect_ocr_trainf — Protection of training data.
Obtaining training data for an OCR classifier can be a complicated and time consuming activity. Thus, training data is valuable and there might be the necessity to protect its usage. The general idea is to prevent protected OCR training data from being reviewed or modified in any way and to ensure that its usage, e.g., training an OCR classifier, is restricted by a user password.
The approach in HALCON is to obtain OCR training data in the usual way, then write it to an unprotected OCR training file using, e.g., write_ocr_trainf, and afterwards create a new file containing the OCR training data protected with a user password. Protected training files can be used with a set of special operators that also require the correct password in order to be able to process the data, e.g., for training an OCR classifier.
It is important to keep in mind that data in a protected OCR training file cannot be reviewed or modified in any way, thus it is important to keep the original, unprotected, training data files.
Protected OCR training data can be used for training of multilayer perceptron (MLP) classifiers and support vector machine (SVM) classifiers.
protect_ocr_trainf protects the OCR training data contained in the file TrainingFile and writes it to the protected training data file TrainingFileProtected. The protection is done with the user password Password. The same password has to be used afterwards in order to use the protected OCR training data, e.g., by the operators trainf_ocr_class_mlp_protected or read_ocr_trainf_names_protected. Empty passwords (or passwords containing the empty string) are not allowed. If the number of passwords Password equals 1, then every output file TrainingFileProtected is protected with that password, otherwise the number of passwords has to be equal to the number of input and output files and the output file at position n is protected with the password at position n.
Names of the training files.
Default value: ''
File extension: .trf, .otr
Passwords for protecting the training files.
Names of the protected training files.
File extension: .trf
If the parameters are correct, the operator protect_ocr_trainf returns the value 2 (H_MSG_TRUE). Otherwise an exception will be raised.
write_ocr_trainf, append_ocr_trainf, concat_ocr_trainf
read_ocr_trainf_names_protected, trainf_ocr_class_mlp_protected, trainf_ocr_class_svm_protected, select_feature_set_trainf_mlp_protected, select_feature_set_trainf_svm_protected